Digital Twin Framework for Aircraft Lifecycle Management Based on Data-Driven Models

Abstract

This paper presents a comprehensive framework for implementing digital twins in aircraft lifecycle management, with a focus on using data-driven models to enhance decision-making and operational efficiency. The proposed framework integrates cutting-edge technologies such as IoT sensors, big data analytics, machine learning, 6G communication, and cloud computing to create a robust digital twin ecosystem. This paper explores the key components of the framework, including lifecycle phases, new technologies, and models for digital twins. It discusses the challenges of creating accurate digital twins during aircraft operation and maintenance and proposes solutions using emerging technologies. The framework incorporates physics-based, data-driven, and hybrid models to simulate and predict aircraft behavior. Supporting components like data management, federated learning, and analytics tools enable seamless integration and operation. This paper also examines decision-making models, a knowledge-driven approach, limitations of current implementations, and future research directions. This holistic framework aims to transform fragmented aircraft data into comprehensive, real-time digital representations that can enhance safety, efficiency, and sustainability throughout the aircraft lifecycle.

1. Introduction

1.1. Background and Motivation

The development of digital twins (DTs) for the operational stage of the aircraft lifecycle presents a unique challenge in the current technological situation. While recent advancements in technologies such as artificial intelligence, Internet of Things (IoT), 6G communications, and edge computing have opened new possibilities for continuous monitoring and data collection, there is a notable absence of comprehensive frameworks to use these technologies for the development of aircraft digital twins (ADTs). This paper addresses this critical gap by proposing a novel framework for implementing digital twins in aircraft lifecycle management.

There is a paradoxical situation in the aviation industry; previously, digital twins could not be built because of technological limitations in continuous monitoring, and now that these technologies are emerging, there is a lack of approaches and models to utilize them effectively. Some of these technologies, such as 6G, are still in experimental stages, adding another layer of complexity to the development of robust digital twin systems.

This article presents not only a theoretical framework but also introduces detailed data-driven, physics-based, and hybrid models designed to simulate and predict aircraft behavior. While empirical validation of the entire framework is challenging because of the current state of the industry, our work provides a necessary foundation for future implementations and studies. The framework is designed to be adaptable and scalable, anticipating the full realization of emerging technologies and setting the stage for practical applications in aircraft lifecycle management.

1.2. Existing Methods and Technologies in ADT

Digital twins, as virtual representations of physical assets, processes, or systems, depend on the integration of real-time data from sensors, simulations, and other sources, as highlighted by Jones et al. [1]. When applied to aircraft structures, this cutting-edge technology enables engineers and decision-makers to monitor and analyze structural behavior comprehensively and continuously throughout the entire lifecycle of the aircraft [2]. Various methodologies have been developed to leverage the data obtained from sensor-equipped structures, turning these insights into valuable information crucial for thorough structural assessments. This integration enhances not only real-time monitoring but also the predictive capabilities of digital models, allowing for a deeper understanding of structural behavior under different conditions. Consequently, the use of sensor data becomes a fundamental element in the evolving practices of structural assessment, supporting a more informed and robust digital-twin approach to ensure the resilience, safety, and optimal performance of critical structures, as proposed in [3].

The application of digital twin concepts in the aeronautical sector was extensively discussed in a position paper by the American Institute of Aeronautics and Astronautics and the Aerospace Industries Association [4]. By replicating the behavior and performance of physical aircraft, digital twins offer a comprehensive means to understand structural responses under diverse operational and environmental conditions, marking significant advancements across multiple dimensions of the aerospace sector, as noted in [5].

Alongside these digital advancements, the development of collaborative tools and cloud-based solutions has further facilitated global teamwork in aircraft design and development [6]. Moreover, advanced simulations, including computational fluid dynamics (CFD) and finite-element analysis (FEA), have evolved to provide precise and comprehensive multi-physical assessments, covering aerodynamics, structural integrity, and thermal performance. Collectively, these innovations have ushered in a new era in aircraft design characterized by heightened efficiency, sustainability, and innovation throughout the aerospace product lifecycle [7].

The integration of artificial intelligence (AI) tools, particularly machine learning (ML), into CFD and FEA presents numerous opportunities for enhancing these fields. In CFD, ML algorithms improve the accuracy and speed of fluid dynamics simulations, enabling more precise modeling of complex flow phenomena [8]. Similarly, in FEA, AI tools, especially machine learning algorithms, offer the potential to enhance prediction accuracy significantly by capturing intricate relationships between input parameters and structural responses. These techniques are invaluable for surrogate modeling, calibration, and updating finite element models [9,10], improving precision in predicting complex non-linearities or uncertainties.

Given the need to enhance the management of aging aircraft and ensure the safety of older aircraft while optimizing operational costs, the article [11] aims to explore the potential benefits of integrating digital twins with model updating techniques in the context of aircraft structure lifecycle simulation. Such integration has the potential to improve current design philosophies within the civil aircraft sector, thereby contributing to the expanding body of knowledge in aircraft structural simulation. The synergy of information from design models and sensor data holds high-impact potential, significantly improving design, analysis, and maintenance processes. This, in turn, enhances overall safety, performance, and cost-effectiveness in aircraft operations, contributing to a more sustainable and environmentally friendly aviation industry [12].

1.3. Related Works

The concept of digital twins has recently emerged as an innovative approach that bridges the physical and digital worlds by creating virtual replicas of physical entities. These digital replicas are employed to simulate, optimize, and predict various scenarios, leading to enhanced efficiency and cost reduction [13]. In the aeroengine sector, digital twin technology is still in its nascent stages. For example, Fentaye et al. [14] proposed a gas path performance diagnostic method that combines support vector machines and artificial neural networks. Lu et al. [15] introduced a diagnostic approach based on Dempster–Shafer evidence theory and data fusion. Additionally, Xiong et al. [16] presented a digital twin framework that integrates data with long short-term memory (LSTM), a deep learning method, for making maintenance decisions for engines. Zhou et al. [17] developed an engine health assessment system by integrating the Kalman Filter with LSTM and incorporating convolutional neural networks. Yanhua et al. [18] created an adaptive correction method for turbofan engines using an LSTM neural network combined with a hybrid optimization method.

However, most of these studies primarily focus on the fusion of data-driven models without fully exploiting the complementary strengths of other mechanisms for digital twin technology [19].

The study [20] proposes a novel digital twin framework that is specifically designed for twin-spool turbofan engines. The proposed digital twin framework combines the advantages of mechanism models and data-driven models to achieve more accurate results.

Zaccaria et al. [21] proposed a comprehensive framework for monitoring, diagnostics, and health management across a fleet of aircraft, utilizing a signature-based algorithm. The digital twin of the associated engine was employed to replicate signatures of various key component faults. Although the primary focus of the paper was physics-based, the authors briefly explored the potential of incorporating neural network classifiers to further enhance the framework. Yang et al. [22] concentrated on the turbofan disk, a critical component prone to crack failures. Their approach to developing a digital surrogate of the disk involved solving existing mathematical models and dynamic equations, adopting a fully physics-based method centered on analyzing vibration response signals to detect crack failures.

Wang et al. [23] emphasized the significance of integrating and fusing heterogeneous data sources to develop a life prediction method for aircraft structures. This involved feeding information such as material properties and structural geometry into a digital twin constructed using finite element methods, specifically for simulating cracks. Chowdhury et al. [24] addressed the environmental control system of an aircraft, deploying their model-based solution directly on a real aircraft during ground testing, demonstrating the practical applicability of their work.

Ezhilarasu et al. [25] focused on health monitoring methods for the aircraft electrical power system. The digital twin of this network, interconnected with other aircraft subsystems, was constructed based on mathematical principles using established simulation tools. The data-driven aspect was represented by an artificial neural network developed for fault isolation and root cause prediction. In a subsequent paper by the same authors [26], they introduced the FAVER framework, which uses DT concepts and reasoning techniques to identify, isolate, and predict faults across interacting aircraft subsystems. Demonstrations of these use cases were presented, showcasing a DT based on both physics and data-driven modeling. This was further elaborated in another article [27], where the authors discussed three use cases, two of which were based on simulation models, and the third involved more efficient hardware-in-the-loop testing. The solutions applied in these studies relied on the open system architecture for condition-based monitoring specification, which facilitated data manipulation algorithms and defined fault classification as part of the state detection and health assessment process.

Ramesh et al. [28] explored aircraft landing gear, utilizing an existing electric braking system model as the DT. By employing a physics-based approach within a software environment, they simulated specific types of faults and incorporated a recurrent neural network to identify these failures accurately.

Huang et al. [29] offered a different perspective on information fusion between physics-based and data-driven models, implementing feed-forward and recurrent neural networks to create an improved DT for real-time fault detection of an aeroengine, specifically by constructing a degradation adaptive correction model. Alvarez et al. [30] tackled the challenge of aircraft airspeed estimation in the event of a pitot tube sensor failure, developing a solution that involved physics-based simulation models combined with data fusion and estimation techniques. Peng et al. [31] centered their project on the aeroengine, conducting practical tests in a test facility to validate their approach.

All major aviation companies have been actively developing platforms to predict component wear and optimize maintenance strategies. Some of the most significant advancements towards digital twins in the aviation sector include Aviatar (Lufthansa Technik, Hamburg, Germany) [32], Skywise (Airbus, Blagnac, France) [33], Predix (General Electric, General Electric. San Ramon, CA, USA) [34], PROGNOS (Air France Industries and KLM Engineering & Maintenance, Paris, France) [35], AnalytX (Boeing, Crystal City, VA, USA) [36], and others. These platforms represent significant strides in the integration of digital twin technologies within the aviation industry, but all of them have the problem of stakeholder borders, which restrict the platforms from gaining the highest possible usage of data.

1.4. Research Gap and Contributions of This Study

Despite the significant advancements in digital twin technology, there is a clear research gap in effectively integrating physics-based models and data-driven models, the two main types of models used in this domain. While each approach has its strengths, existing frameworks tend to focus on one or the other, missing the opportunity to use the complementary advantages of both.

Physics-based models rely on established engineering principles to simulate the physical behavior of systems. These models are particularly effective at capturing structural dynamics, aerodynamics, and other well-understood phenomena, but they can be computationally expensive and less adaptable to real-time changes.

On the other hand, data-driven models utilize machine learning and statistical methods to predict outcomes based on historical and real-time data. These models are highly adaptable and capable of improving through experience, but they often lack the precision and deep understanding of the physical processes that physics-based models provide.

The research gap lies in the lack of a robust framework that can seamlessly integrate these two types of models—combining the accuracy and reliability of physics-based models with the flexibility and adaptability of data-driven models. This integration is crucial for creating digital twins that not only simulate aircraft systems accurately but also update dynamically in real time, based on sensor data and operational conditions.

The key challenge in integrating these models arises from the complexity of ensuring continuous monitoring and the need for real-time updates. Existing frameworks either fail to account for this continuous feedback loop or do not address the challenges of integrating real-time data into physics-based simulations. Furthermore, the current state of research lacks methods to handle vast amounts of heterogeneous data securely and efficiently across different platforms, systems, and stakeholders.

This paper addresses the identified gap by proposing a novel framework that integrates both physics-based and data-driven models for aircraft lifecycle management. By using emerging technologies such as IoT, AI, 6G, and federated learning, this framework provides a scalable and secure solution for updating digital twins in real time. The proposed approach allows for more accurate predictions and simulations, paving the way for the next generation of digital twin applications in aviation.

1.5. Paper Structure

This paper is structured as follows: Section 2 presents the comprehensive digital twin framework for aircraft lifecycle management, detailing its key components including lifecycle phases, new technologies, models for digital twins, and supporting components. Section 3 discusses the results of implementing this framework, addressing challenges in creating digital twins during aircraft operation, presenting the architecture of the ADT ecosystem, and exploring each component of the framework in depth. Section 4 provides a discussion on decision-making processes within the ADT framework, introduces a knowledge-driven approach, examines the challenges and limitations in implementing ADT, and outlines future research directions. Section 5 concludes this paper, summarizing the key contributions and implications of this work for the future of aircraft lifecycle management.

2. Study Framework for Aircraft Lifecycle Management Based on Data-Driven Models

2.1. Framework Development as the Core Methodology

The development of the digital twin framework for aircraft lifecycle management constitutes the core methodology of this research. This approach aligns with the design science research paradigm, where the creation of novel artifacts—in this case, our comprehensive framework—is itself a primary research contribution. Our methodology involves the synthesis of existing technologies, theoretical concepts, and industry needs into a cohesive, actionable framework. This process includes the following:

• Identification and analysis of key components necessary for an effective aircraft digital twin.
• Integration of cutting-edge technologies such as IoT, AI, and 6G into the framework.
• Development of novel models for data processing, simulation, and decision-making within the digital twin ecosystem.
• Design of supporting components to ensure the framework’s practical implementation.

The following subsections detail this framework, which serves as both our research methodology and a key contribution to the field of aircraft lifecycle management.

2.2. Digital Twin Framework

The concept of the digital twin has emerged as a transformative tool for managing the lifecycle of aircraft. A DT is a dynamic, data-driven virtual representation of a physical aircraft, capable of mirroring its real-world counterpart in real time.

The digital twin framework for aircraft lifecycle management based on data-driven models is a holistic approach that integrates various components and phases into a cohesive system for managing the entire lifecycle of an aircraft. This framework is depicted as a triangular structure with three primary vertices as follows: lifecycle phases, new technologies, and models for digital twins. Additionally, there are supporting components, which facilitate the interaction and integration of these elements. Figure 1 illustrates this framework, emphasizing the interconnectedness of each component.

This article explores the digital twin framework for aircraft lifecycle management, focusing on the integration of data-driven models and the latest technological advancements.

It is important to highlight that the approach described in this manuscript fundamentally differs from previous research, particularly from the work discussed in [37], which focused on leveraging artificial intelligence of things (AIoT) technologies for individual aircraft health monitoring. While [37] presents a promising framework for enhancing aviation health monitoring systems through AIoT, the focus was primarily on single-instance aircraft health data collection and analysis. The current study advances this concept by proposing an ecosystem-wide digital twin architecture. This architecture integrates not only real-time data from individual aircraft but also data from the entire aircraft fleet, leveraging federated learning and new interaction models facilitated by an AIoT platform.

A critical difference between the two approaches is the capacity to model the lifecycle of the entire fleet, including stakeholder transitions and environmental changes. For instance, the digital twin ecosystem can account for factors such as an aircraft being relocated from a hot, arid climate to a cold, humid region, or a change in ownership that could lead to a drastic shift in maintenance regimes. Such considerations are essential for improving long-term operational efficiency and aircraft safety. Moreover, these changes are not limited to individual aircraft but span the entire fleet, requiring a more sophisticated, adaptive ecosystem architecture.

The need for detailed architectural development and interaction models within the aviation twin ecosystem is underscored by the complexity and dynamism of modern aviation operations. Developing robust architectures that support predictive maintenance, real-time adjustments, and long-term optimization across the fleet is not only necessary but also crucial for the next generation of aviation health monitoring systems. Therefore, the approaches outlined in this manuscript go beyond the scope of the existing frameworks and offer a novel contribution to the digital twin ecosystem for aviation, highlighting the importance of building detailed models and interaction protocols to support these advanced systems.

2.3. Lifecycle Phases

The lifecycle of an aircraft and its corresponding digital twin are closely intertwined, with the phases of the aircraft’s life being mirrored by the phases of the digital twin’s lifecycle. However, while these phases coincide in time, they differ in terms of the specific tasks and objectives being pursued.

During the design and development phase of the aircraft, the digital twin is initially created. This involves integrating detailed design specifications, engineering data, and simulations into a comprehensive virtual model. The digital twin during this phase serves as a dynamic prototype, allowing engineers to test various design scenarios, optimize performance parameters, and foresee potential issues before physical production begins.

In the manufacturing phase, the aircraft is constructed according to the design specifications. Simultaneously, the digital twin is updated to reflect the “as-built” condition of the aircraft, incorporating real-time data from the production process. This ensures that the digital twin accurately represents the physical aircraft, including any deviations from the original design that occur during manufacturing.

Once the aircraft is operational, the digital twin enters its most active phase. It continuously receives data from sensors embedded in the aircraft, as well as from external sources such as environmental conditions and operational feedback. These real-time data are used to monitor the aircraft’s performance, predict maintenance needs, and optimize operational efficiency. The digital twin becomes a vital tool for ensuring that the aircraft operates safely and efficiently throughout its service life.

Throughout the aircraft’s operational life, it undergoes regular maintenance, repair, and overhaul (MRO). The digital twin plays a crucial role in this phase by providing accurate predictions of when and where maintenance is needed. It uses historical data, real-time monitoring, and predictive analytics to schedule maintenance activities, thus minimizing downtime and extending the aircraft’s service life. The digital twin is also updated with new information each time maintenance is performed, ensuring it remains an accurate reflection of the aircraft’s current condition.

As the aircraft approaches the end of its lifecycle, the digital twin helps manage the decommissioning process. It provides valuable insights into which components can be recycled or reused and assists in planning the dismantling process. The data accumulated over the aircraft’s life can be used to inform decisions about future designs and operations, making the digital twin a lasting resource even after the physical aircraft is retired.

2.4. New Technologies

The successful implementation of the DT framework relies heavily on several cutting-edge technologies. Internet of Things (IoT) devices and sensors are integral to the real-time data collection process, providing continuous updates to the digital twin from the aircraft’s various systems. The data collected by these sensors are critical for maintaining an accurate and up-to-date digital twin throughout the aircraft’s lifecycle.

AI and ML are equally important in the framework, as they enable the analysis of vast amounts of data and the training of predictive models. These models help in anticipating maintenance needs, optimizing operational performance, and detecting anomalies before they lead to serious issues. AI and ML allow the digital twin to learn from historical and real-time data, improving its accuracy and predictive capabilities over time.

The framework also leverages 6G connectivity to ensure seamless communication between the aircraft, data centers, and the digital twin. The high-speed, low-latency capabilities of 6G networks enable the rapid transmission of large volumes of data, ensuring that the digital twin is always current and reflective of the aircraft’s actual condition. This connectivity is particularly important for real-time monitoring and decision-making, where delays could have significant consequences.

Blockchain technology provides the security and authenticity required for the data integrated into the digital twin, especially in the context of maintenance records. By securing these records on a blockchain, the framework ensures that the history of the aircraft’s maintenance activities is preserved and cannot be tampered with, even as the aircraft changes ownership or undergoes repairs by different service providers.

2.5. Models for Digital Twins

At the heart of the DT framework are the models that simulate and predict the behavior of the aircraft. Physics-based models form the foundation of the digital twin, simulating the physical behavior of the aircraft based on its design and engineering parameters. These models include structural dynamics, aerodynamics, and system behaviors, providing a detailed and accurate representation of how the aircraft is expected to perform under various conditions.

Data-driven models complement the physics-based models by using historical data to predict future performance and maintenance needs. Machine learning algorithms, such as neural networks and ensemble methods, are trained on large datasets to identify patterns and trends that can inform predictive maintenance and operational optimization. These models are particularly valuable in the operation phase, where real-time data are continuously fed into the digital twin.

Hybrid models integrate the strengths of both physics-based and data-driven approaches to create a more comprehensive digital twin. These models use techniques such as Kalman filters to combine real-time data with physical simulations, providing a more accurate and dynamic representation of the aircraft’s current state. Hybrid models are essential for ensuring that the digital twin remains a reliable tool for decision-making throughout the aircraft’s lifecycle.

2.6. Supporting Components

The supporting components of the DT framework are crucial for ensuring that the lifecycle phases, technologies, and models work together seamlessly. Data management and integration platforms are responsible for the robust handling of data from various sources, ensuring that they flow smoothly into the digital twin. These platforms include data pipelines, storage solutions, and processing tools that handle the vast amounts of data generated throughout the aircraft’s lifecycle.

Federated learning and collaboration tools facilitate collaboration among the various stakeholders involved in the aircraft’s lifecycle, such as airlines, MRO providers, and OEMs. Federated learning allows these stakeholders to contribute to the digital twin’s development without sharing sensitive data, ensuring privacy and security. The collaboration tools also include governance frameworks that define the rules for data sharing and model updates, ensuring that all parties adhere to the agreed-upon protocols.

Analytics and decision support tools provide the insights needed for informed decision-making. These tools analyze the data collected and the models developed within the digital twin to generate actionable recommendations. Whether optimizing flight paths, scheduling maintenance, or detecting anomalies, these tools ensure that the digital twin is not just a passive representation but an active participant in the aircraft’s management.

3. Results

3.1. The Challenge of Creating a Digital Twin of an Aircraft during Technical Operation and Maintenance

In the modern aerospace industry, the design and production of aircraft have become entirely digitized, creating a seamless and precise foundation for building a digital twin—a virtual representation that mirrors the physical aircraft in real time. This digitalization in the early stages of an aircraft’s lifecycle ensures that every component, system, and specification is meticulously documented and can be modeled with high fidelity. However, as the aircraft moves beyond production and into the operational phase, creating and maintaining an accurate digital twin becomes increasingly complex. This challenge is primarily due to the multitude of variables and unpredictable factors that emerge during the aircraft’s operational life, which have been difficult or even impossible to account for—until now.

One of the primary challenges in maintaining an accurate digital twin during the operational phase is the variability in individual aircraft configurations and operating environments. Although an aircraft may begin its service life identical to its sister models, it quickly diverges as it undergoes modifications and experiences unique operational conditions. For instance, individual changes in an aircraft’s design and systems, made to adapt to specific operating modes, can significantly alter its characteristics. These modifications may be driven by factors such as geographic location, where variations in temperature, humidity, sand, wind, and other environmental conditions can have substantial effects on the aircraft’s structure and systems. The operational environment, whether it involves extreme cold or intense heat, influences how the aircraft’s materials behave, leading to changes that a generic digital model cannot predict without specific data inputs.

Furthermore, the commercial operation of an aircraft introduces additional complexities. The number of hours an aircraft spends in the air, the usage patterns of engines during taxiing, landing techniques, and even the quality of runway surfaces all contribute to the unique wear and tear on each aircraft. These factors are not only difficult to predict but also vary widely from one aircraft to another, making it challenging to create a digital twin that accurately reflects the real-world condition of each individual aircraft.

The difficulties extend beyond the physical wear and operational variables to the domain of maintenance and documentation. Aircraft maintenance is a continuous process, and the history of this maintenance is crucial for understanding the current state of the aircraft. However, in practice, this documentation is often fragmented. Despite advancements in digital record-keeping, a significant portion of maintenance documentation remains in handwritten form or stored in disparate computer systems. When an aircraft changes ownership—a common occurrence over its lifespan—its maintenance history often does not transfer in a complete or coherent form. This loss of historical data effectively resets the understanding of the aircraft’s condition, further complicating the creation of a continuous and accurate digital twin.

Adding to the complexity is the fact that maintenance can be performed by different organizations throughout the aircraft’s life. Each organization may have its own methods for recording and storing data, leading to inconsistencies and gaps in the maintenance history. These gaps are critical because they can obscure the detection of long-term trends and the identification of systemic issues that could impact the aircraft’s safety and performance. Without a comprehensive and continuous maintenance record, the digital twin cannot accurately reflect the true state of the aircraft, limiting its utility in predicting future failures or optimizing maintenance schedules.

Given these challenges, the creation of a digital twin during the operational and maintenance stages of an aircraft’s lifecycle has been almost impossible. The absence of a cohesive and consistent data flow from the time the aircraft enters service to its retirement means that any virtual model would be incomplete or inaccurate. However, the emergence of new technologies is beginning to change this landscape. Advances on IoT, big data analytics, and AI are enabling more comprehensive data collection and analysis, even in the complex and variable environment of aircraft operation.

Sensors embedded throughout the aircraft can now collect data continuously, providing real-time insights into the aircraft’s condition and performance. These data can be fed into a digital twin, allowing it to evolve alongside the physical aircraft. Moreover, blockchain technology offers a potential solution for maintaining a secure and immutable record of maintenance history, ensuring that all relevant data are preserved and accessible, regardless of ownership changes. Additionally, machine learning algorithms can analyze these data to detect patterns and predict future maintenance needs, enhancing the accuracy and usefulness of the digital twin.

3.2. Architecture of the Aircraft Digital Twin Ecosystem

The architecture of the aircraft digital twin ecosystem represents a transformative approach to aviation management and operations, integrating cutting-edge technologies to create a comprehensive, interconnected system that spans multiple domains and stakeholders. The architecture of the ADT ecosystem (Figure 2) integrates various components into a cohesive system that enables real-time monitoring, predictive maintenance, and optimization of aircraft operations. The key elements of this architecture include the ADT ecosystem, artificial intelligence of things (AIoT) platform, federated learning, and 6G connectivity. Below is a detailed description of each component and their interconnections.

• Aircraft digital twin ecosystem.

The ADT ecosystem consists of a network of stakeholders and components that contribute to the creation, updating, and utilization of the digital twin. This ecosystem includes the following:

• Aircraft health monitoring systems (AHMSs) with sensors embedded within the aircraft continuously collect data on various parameters such as structural integrity, engine performance, and environmental conditions. These data are crucial for updating the digital twin in real time.
• The digital twin is a virtual replica of the physical aircraft, continuously updated with data from the AHMS and other sources. It provides a real-time representation of the aircraft’s condition and is used for predictive maintenance, performance optimization, and operational planning.
• Aircraft manufacturers provide the initial digital model and design specifications that serve as the foundation for the digital twin. They also contribute data on component performance and manufacturing processes.
• Original equipment manufacturer (OEM) suppliers contribute data on specific parts and systems they produce, which are integrated into the digital twin to reflect the detailed behavior and lifecycle of each component.
• Airlines operate the aircraft and generate vast amounts of operational data, including usage patterns, flight hours, and maintenance records. These data are fed into the digital twin to enhance its accuracy and predictive capabilities.
• MRO providers are responsible for the upkeep and repair of the aircraft. They contribute data on maintenance activities, which is essential for maintaining an accurate digital twin and predicting future maintenance needs.
• Airports provide data on environmental conditions, runway usage, and other operational factors that affect the aircraft. This information is used to update the digital twin with relevant contextual data.
• Air traffic control (ATC) systems provide data on flight paths, airspace conditions, and other factors that impact the aircraft during flight. These data are integrated into the digital twin for real-time operational insights.

2.

AIoT platform.

The AIoT platform is a central component that integrates data from the ADT ecosystem and performs advanced analytics and machine learning tasks. Its key features include the following:

• Data aggregation by the AIoT platform, which collects data from various sources within the ADT ecosystem, including AHMS, MROs, airlines, and more.
• Advanced analytics integrated into the platform, which use AI and machine learning algorithms to analyze the aggregated data, identify patterns, predict failures, and optimize operational efficiency.
• Model training by the AIoT platform, which is also responsible for training machine learning models that are used within the digital twin to improve its predictive capabilities.

3.

Federated learning.

Federated learning is a decentralized machine learning approach that allows multiple stakeholders within the ADT ecosystem to train models collaboratively without sharing raw data. This approach is critical in the aviation industry because of data privacy concerns and the competitive nature of the industry. Its key aspects include the following:

• Model updates in the process when each stakeholder trains a local model using its own data. The model updates (gradients) are then shared with a central server or aggregation mechanism.
• Central aggregation is the process where the central server aggregates these updates to create a global model, which is then distributed back to the stakeholders.
• Since raw data never leave the local environment, federated learning ensures that sensitive information remains secure while still benefiting from collaborative learning.

4.

The role of 6G connectivity.

The ADT ecosystem relies on 6G connectivity to ensure seamless and real-time data transmission between the aircraft, AIoT platform, and various stakeholders. The features of 6G include the following:

• Ultra-fast data transfer rates are provided by 6G, which are necessary for transmitting large volumes of data generated by AHMS and other sensors in real time.
• The low latency of 6G networks ensures that data are transmitted with minimal delay, enabling real-time updates of the digital twin and rapid response to emerging issues.
• Several 6G networks, including satellite communication across different orbits (LEO, MEO, GEO), provide continuous connectivity even in remote or high-altitude regions, ensuring that data flow uninterruptedly between the aircraft and the AIoT platform, essential for continuous data transmission in areas where terrestrial networks may not be available. This includes UAVs, airships, and high-altitude platforms that enhance connectivity and provide additional data collection capabilities, ensuring that the digital twin is updated with the latest information from the aircraft.

5.

Cloud data center.

The cloud data center serves as the centralized location for data storage, processing, and model aggregation. It supports the AIoT platform by providing the computational resources needed for real-time data processing, machine learning, and model distribution.

The architecture of the ADT ecosystem, as depicted in Figure 2, integrates a wide range of technologies and stakeholders into a cohesive system that enhances the safety, efficiency, and operational capabilities of aircraft. By using AIoT platforms, federated learning, and 6G connectivity, the ecosystem ensures that the digital twin is always accurate and up to date, providing real-time insights that enable predictive maintenance, optimized performance, and proactive decision-making across the entire lifecycle of the aircraft.

3.3. Distinguishing between the Aircraft Lifecycle and the Digital Twin Lifecycle

In the modern aviation industry, digital twin technology has emerged as a transformative tool for optimizing the lifecycle management of aircraft. However, it is crucial to distinguish between the following intertwined but distinct concepts: the lifecycle of the aircraft and the lifecycle of the digital twin. While these two cycles are closely related, they serve different purposes and evolve under different sets of rules. Understanding the distinction between these lifecycles is essential to grasp the full benefits and limitations of digital twin technology in aviation.

• The Aircraft Lifecycle.

The aircraft lifecycle refers to the various stages that an aircraft undergoes from its conceptual design to its retirement and decommissioning. This lifecycle is typically divided into five key phases as follows: design, manufacturing, operation, maintenance, and decommissioning. Each phase requires meticulous planning, resource allocation, and decision-making to ensure the safe, efficient, and profitable operation of the aircraft.

Design Phase. The design phase is the first step in the lifecycle, where engineers and designers create the blueprint for the aircraft. Factors such as aerodynamics, materials, safety, and regulatory compliance are considered, with simulations and physical prototypes developed to test various design concepts. The aim of this phase is to ensure that the aircraft is designed to meet the desired performance standards. Use case: Boeing employs digital twins to evaluate design choices for new aircraft models, enabling virtual testing of thousands of design variables to optimize fuel efficiency and structural integrity without the need for physical prototypes [38].

Manufacturing Phase. Once the design is finalized, the manufacturing phase involves the actual production of the aircraft. This phase includes assembling the aircraft’s components and ensuring that every part meets the specifications set during the design phase. Quality control is critical in this phase to prevent defects that could impact safety or performance. Use case: Airbus uses digital twin technology to monitor its assembly processes, ensuring that every component is precisely manufactured and assembled according to design specifications, thereby reducing errors and improving efficiency [39].

Operational Phase. During the operational phase, the aircraft is in active use. It is flown for commercial, military, or private purposes, depending on its design. Operational efficiency, fuel consumption, safety, and performance monitoring become key concerns during this period, as the aircraft must meet its intended performance benchmarks under real-world conditions. Use case: GE Aviation uses digital twins for real-time engine performance monitoring, helping airlines optimize fuel efficiency while predicting maintenance needs to avoid costly in-flight failures [40].

Maintenance Phase. Aircraft require regular inspections, repairs, and updates to ensure continued safety and performance. Maintenance includes both scheduled activities, like routine inspections and part replacements, and unscheduled activities, such as repairs following unexpected failures. Effective maintenance strategies are crucial for extending the lifespan of the aircraft and minimizing downtime. Use case: Rolls-Royce’s digital twin models for jet engines predict when specific parts will require maintenance, allowing airlines to schedule repairs in advance and avoid unscheduled downtime [41].

Decommissioning Phase. The final phase of the aircraft lifecycle involves retiring the aircraft from active service. This may include the disassembly and recycling of components or the complete disposal of the aircraft. Decommissioning decisions often depend on factors such as the aircraft’s condition, the availability of newer technology, and the costs associated with continued operation. Use case: Aircraft recycling companies use digital twins to assess which components of decommissioned aircraft, such as engines and avionics, can be refurbished and reused in other applications [42].

2.

The Digital Twin Lifecycle.

The digital twin lifecycle, while aligned with the aircraft’s lifecycle, operates as a dynamic, data-driven replica of the physical aircraft, evolving alongside its real-world counterpart. Unlike the aircraft lifecycle, which is primarily concerned with physical processes, the digital twin lifecycle is driven by continuous data collection, simulation, and analysis. The digital twin is not a static entity but a dynamic system that evolves through different stages to reflect the condition, performance, and behavior of the physical aircraft in real time.

Creation Phase. The digital twin is first created during the design phase of the aircraft. It starts as a virtual model, incorporating the same design specifications, simulations, and operational requirements as the physical aircraft. At this stage, the digital twin is used primarily for testing and optimization, allowing engineers to simulate different scenarios and identify potential issues before the aircraft is built.

Integration Phase. As the aircraft moves into the manufacturing phase, the digital twin is updated to reflect the “as-built” condition of the aircraft. Real-time data from the production process are fed into the digital twin, ensuring that any deviations from the original design are captured. This ensures that the digital twin remains an accurate representation of the physical aircraft as it transitions from design to reality.

Operational Phase. In the operational phase, the digital twin plays its most active role. It continuously collects and processes data from sensors embedded in the aircraft, providing real-time insights into the performance and condition of the aircraft. This allows for predictive maintenance, operational optimization, and the ability to simulate “what-if” scenarios to prevent failures or optimize fuel consumption. The digital twin evolves as new data are fed into it, making it a dynamic tool for decision-making.

Maintenance Support Phase. During the maintenance phase of the aircraft lifecycle, the digital twin becomes a critical tool for predicting maintenance needs and scheduling repairs. By analyzing historical data and real-time inputs, the digital twin can forecast when specific components are likely to fail, helping to reduce unplanned downtime and improve safety. The digital twin is updated after every maintenance event, ensuring that it accurately reflects the aircraft’s current condition.

Decommissioning Phase. Just as the aircraft lifecycle ends with decommissioning, the digital twin also enters a final phase. During decommissioning, the digital twin provides valuable insights into the condition of various components and systems, helping operators decide which parts can be reused, refurbished, or recycled. Even after the physical aircraft is decommissioned, the digital twin can continue to serve as a valuable resource for historical analysis and future design improvements.

3.

Key Differences in the Two Lifecycles.

While the aircraft lifecycle and the digital twin lifecycle are closely intertwined, they differ in the following key aspects:

• Physical vs. Virtual. The aircraft lifecycle deals with the tangible, physical aspects of the aircraft’s existence, from its construction to its eventual dismantling. In contrast, the digital twin lifecycle is virtual, driven by data and simulation rather than physical processes. The digital twin mirrors the aircraft, but it exists in a digital realm.
• Static vs. Dynamic. The aircraft lifecycle progresses through relatively static stages, each marked by discrete events such as production, operation, and maintenance. The digital twin, however, is a dynamic entity that evolves continuously throughout the aircraft’s lifecycle. It adapts to new data and changing conditions, making it a real-time tool for analysis and decision-making.
• Creation Timing. The digital twin is created in tandem with the aircraft but evolves more rapidly during the early phases. It provides real-time insights from the moment the aircraft is manufactured and continues to update and evolve with every new data input. The aircraft itself, by contrast, remains in a relatively fixed state during certain phases, particularly in operation and maintenance.
• Utility Beyond Decommissioning. While the aircraft lifecycle typically ends at decommissioning, the digital twin may continue to provide value after the aircraft has been retired. The data stored in the digital twin can be used for post-decommissioning analysis, informing the design and operational strategies of future aircraft models.

4.

The Interplay Between Lifecycles.

Although distinct, the lifecycles of the aircraft and its digital twin are deeply interconnected. The digital twin’s role is to enhance and optimize each phase of the aircraft’s lifecycle by providing real-time data, predictive analytics, and simulations that can be used to improve decision-making at every stage. As the aircraft evolves through its lifecycle, the digital twin mirrors these changes, ensuring that operators have access to accurate, up-to-date information. In this sense, the digital twin can be seen as a tool that enhances the effectiveness of the aircraft lifecycle by reducing risks, minimizing downtime, and improving overall performance.

3.4. Lifecycle Phases as a Component of the ADT Framework

The first component of the ADT framework for aircraft lifecycle management (Figure 1) is DT lifecycle phases on the base of a comprehensive approach that integrates multiple technologies, methodologies, and processes across the entire lifecycle of the aircraft. This approach ensures that the digital twin is accurately created, continuously updated, and effectively utilized for predictive maintenance, operational optimization, and decision-making. This framework has some key phases and components.

Let us examine them in more detail, using the notations provided in

Table 1. Notations and definitions.

Symbols	Definition
$\theta$	Model parameter that represents the set of parameters that define the digital twin’s behavior, such as structural properties, aerodynamic coefficients, and system performance metrics.
${\mathcal{M}}_0$	Initial digital twin model created during the design phase, incorporating physics-based models, data-driven models, and initial conditions.
${\mathcal{M}}_{struct}$	The structural model of the digital twin. It is a specific model that focuses on simulating the structural dynamics and behavior of the aircraft. This model includes aspects such as how the aircraft’s frame and materials respond to various forces, stresses, and environmental conditions.
${\mathcal{M}}_{aero}$	The aerodynamic model of the digital twin. It is a specific model that focuses on simulating the aerodynamic behavior of the aircraft. This model would typically include calculations related to how air flows around the aircraft, the forces generated by this airflow (like lift and drag), and how these forces affect the aircraft’s performance.
${\mathcal{M}}_{systems}$	The systems model of the digital twin. It focuses on simulating the behavior and performance of various onboard systems of the aircraft, such as the propulsion system, avionics, electrical systems, hydraulic systems, and other critical components that ensure the aircraft’s functionality.
${\mathcal{M}}_{data}$	The data-driven model within the digital twin framework. This model is based on empirical data and statistical or machine learning methods rather than purely physics-based equations. It is used to predict the behavior of the aircraft by learning from historical data patterns.
${\mathcal{X}}_{srtuct}$	The input features or conditions relevant to the structural model. These inputs could include material properties (such as Young’s modulus, Poisson’s ratio, and material density), geometric properties (such as dimensions of the aircraft’s components, thickness of materials, and cross-sectional areas), loading conditions (such as external forces, moments, pressure distributions, and thermal loads), and boundary conditions (such as fixed supports, constraints, and connections between components).
${\mathcal{X}}_{aero}$	The input features or conditions relevant to the aerodynamic model. These inputs could include flight conditions (such as airspeed, altitude, angle of attack, and Mach number), aircraft geometry (such as the shape and size of the wings, fuselage, and control surfaces), environmental conditions (such as air density, temperature, and atmospheric pressure), and control inputs (such as deflections of control surfaces (e.g., ailerons, elevators, rudders)).
${\mathcal{X}}_{systems}$	The input features or conditions relevant to the systems model. These inputs could include system configurations (such as engine settings, hydraulic pressures, and electrical loads), control inputs (such as throttle position, autopilot settings, and actuator commands), environmental conditions (such as temperature, altitude, and external forces that might impact system performance), and operational condition (such as flight phase and power demands).
${\mathcal{X}}_{historical}$	The historical data inputs that are used to train or inform the data-driven model. These inputs might include past operational data (such as flight logs, performance metrics, and environmental conditions experienced by the aircraft over time), maintenance records (information on past repairs, inspections, and component replacements), failure data (instances of system or component failures and the conditions under which they occurred), and usage patterns (data on how the aircraft has been used over time, including flight hours, routes, and load factors).
${\mathcal{D}}_t$	Real-time data at time $t$—the set of data collected from sensors and systems at a specific time ttt, including operational and environmental conditions.
${\mathit{x}}_t$	Feature vector at time $t$—a vector representing the input features or conditions observed at time ttt, such as sensor readings, environmental factors, and operational parameters.
${\mathit{y}}_t$	Target variable at time $t$—the observed output or target variable at time ttt, such as system performance metrics, component wear levels, or other measurable outcomes.
$\mathcal{L}(\theta )$	The loss function is a function that measures the difference between the predicted outcomes and the actual observed outcomes, used to optimize the model parameters.
${\nabla }_{\theta }\mathcal{L}(\theta )$	Gradient of the loss function with respect to $\theta$, which represents the direction and rate of change in the loss function with respect to the model parameters $\theta$, used in optimization algorithms.
$\eta$	Learning rate—a scalar value that controls the step size in the gradient descent optimization process, determining how much the model parameters are adjusted at each iteration.
${\Delta \theta }_k^{(t)}$	Local model update for stakeholder $k$ at time $t$—the change in model parameters computed by stakeholder $k$ based on local data and used in federated learning to update the global model.
${\theta }^{(t)}$	Global model parameters at time $t$—the set of model parameters that define the state of the digital twin at a specific time $t$, updated continuously through federated learning and real-time data integration.
$T_f$	The time to failure of the component or system. It is a random variable that represents the point in time when the failure is expected to occur.
${\hat{T}}_f$	Predicted time to failure—the estimated time at which a component or system is expected to fail, based on current operational conditions and historical data.
${\mathcal{O}\{\theta }^{(t)}, {\mathit{x}}_t\}$	The objective function is a function representing the goal of optimization (e.g., minimizing fuel consumption, maximizing performance) as a function of the model parameters and current conditions.
${\mathcal{C}}_{maint}({\hat{T}}_f)$	The cost function for maintenance is a function that represents the cost associated with maintenance activities, depending on the predicted time to failure and other factors.
${\mathcal{C}\{\theta }^{(t)}, {\mathit{x}}_t\}$	Constraints—a set of conditions or limits (e.g., safety regulations, operational limits) that must be satisfied during optimization.
${‖{\mathit{y}}_t-f\{{\mathit{x}}_t;{\theta }^{(t)}‖}^2$	Anomaly score—the squared difference between the observed outcomes and the predicted outcomes, used to detect deviations from normal behavior.

.

• Design Phase

The design phase involves the creation of the initial digital twin, which represents the aircraft’s design, manufacturing details, and expected operational performance. This phase can be broken down into the following steps:

(a)

Model Definition

Physics-based modeling starts by defining the physical and mechanical properties of the aircraft. This includes structural models (e.g., finite element models), aerodynamic models (e.g., CFD simulations), and system models (e.g., electrical, hydraulic systems):

$${\mathcal{M}}_{struct}=f_{struct}({\mathcal{X}}_{srtuct})$$

$${\mathcal{M}}_{aero}=f_{aero}({\mathcal{X}}_{aero})$$

$${\mathcal{M}}_{systems}=f_{systems}({\mathcal{X}}_{systems})$$

where

• $f_{struct}$ denotes the function or algorithm that defines how the structural model processes the input variables. It could involve a set of equations or a computational method (e.g., finite element analysis) that describes the physical principles governing the aircraft’s structural behavior.
• $f_{aero}$ denotes the function or algorithm that defines how the aerodynamic model processes the input variables. This could involve a set of equations, such as those derived from Navier–Stokes equations, or a computational method like computational fluid dynamics that simulates the airflow and aerodynamic forces acting on the aircraft.
• $f_{systems}$ denotes the function or algorithm that defines how the systems model processes the input variables. This function models the interactions and behaviors of the aircraft’s systems based on physical laws, system dynamics, and control algorithms.

Data-driven modeling incorporates historical data and empirical models derived from previous aircraft designs and operations. This includes initial predictive model, which could involve statistical models or initial machine learning models trained on historical datasets:

$${\mathcal{M}}_{data}=g({\mathcal{X}}_{historical})$$

where

• $g$ denotes the function or algorithm that processes historical data to build the data-driven model. The function $g$ could be a machine learning algorithm such as a neural network, regression model, decision tree, or any other statistical model that can learn from data.

(b)

Digital twin creation

The digital twin is created by integrating the physics-based and data-driven models into a comprehensive initial digital twin model ${\mathcal{M}}_0$. This model should capture all essential aspects of the aircraft’s expected behavior under various scenarios:

$${\mathcal{M}}_0={\mathcal{M}}_{struct}\cup {\mathcal{M}}_{aero}\cup {\mathcal{M}}_{systems}\cup {\mathcal{M}}_{data}$$

Simulations are executed to validate the digital twin against expected performance metrics. Then, the model is corrected and modified based on the simulation results to ensure its accuracy.

$${{\mathcal{Y}}_{sim}=\mathcal{M}}_0({\mathcal{X}}_{sim})$$

where

• ${\mathcal{Y}}_{sim}$ represents the simulated outputs or results. It is the predicted behavior or performance metrics generated by the initial digital twin model ${\mathcal{M}}_0$ when it processes the input conditions. Examples of ${\mathcal{Y}}_{sim}$ might include predicted stress levels in the structure, estimated aerodynamic forces, or expected system performance under specific conditions.
• ${\mathcal{X}}_{sim}$ refers to the input features or conditions used in the simulation. These are the specific variables or parameters that are fed into the digital twin model to simulate the aircraft’s behavior. ${\mathcal{X}}_{sim}$ could include factors like the aircraft’s speed, altitude, angle of attack, environmental conditions (such as temperature and pressure), and other relevant operational inputs.

2.

Support Phase

The support phase involves the continuous updating and refinement of the digital twin as the aircraft operates. This phase is crucial for maintaining the accuracy and relevance of the digital twin.

(a)

Real-time data integration

Real-time data are collected from aircraft health monitoring systems, which include sensors monitoring various aircraft parameters such as structural loads, engine performance, and environmental conditions.

$${\mathcal{D}}_t=\{{\mathit{x}}_t,{\mathit{y}}_t$$

The data are cleaned, normalized, and preprocessed to ensure they are in a suitable format for model updates. This step might involve filtering noise, handling missing values, and normalizing data across different sensors and systems:

$${\mathcal{D}\prime }_t=Preprocess({\mathcal{D}}_t)$$

where ${\mathcal{D}\prime }_t$ represents the preprocessed data at time $t$. After applying the preprocessing function to the raw data ${\mathcal{D}}_t$, the result is a cleaned, normalized, and ready-to-use dataset ${\mathcal{D}\prime }_t$ that can be fed into machine learning models, simulations, or further analysis.

(b)

Model updates and federated learning

The preprocessed data are used to update and train local models at each stakeholder site (e.g., airlines, MRO providers). These updates are performed using techniques like stochastic gradient descent:

$${\Delta \theta }_k^{(t)}=\eta {\nabla }_{\theta }{\mathcal{L}}_k\left\{{\theta }^{\left(t\right)}\right\}$$

(1)

Federated learning is used to aggregate local model updates from various stakeholders to refine the global digital twin model. Federated learning ensures that sensitive data remains with the stakeholders while still contributing to the global model:

$${\theta }^{\left(t+1\right)}={\theta }^{\left(t\right)}+\sum _{k=1}^K{\displaystyle \frac{\left|{\mathcal{D}}_k\right|}{\left|\mathcal{D}\right|}}{\Delta \theta }_k^{(t)}$$

(2)

The model is continuously refined as new data come in to ensure that the digital twin evolves to reflect the current state of the aircraft:

$${\theta }^{\left(t+1\right)}{=\theta }^{\left(t\right)}-{\mathsf{\eta }\nabla }_{\theta }{\mathcal{L}}_k\left\{{\theta }^{\left(t\right)},{\mathcal{D}}_t\right\}$$

(3)

The updated global model ${\theta }^{(t+1)}$ is distributed back to all stakeholders, who use it as the starting point for their next round of local training:

$${\theta }_k^{(t+1)}={\theta }^{\left(t+1\right)}, \forall k=\overline{1,K}$$

(4)

This ensures that all stakeholders benefit from the collective knowledge while maintaining data privacy.

3.

Usage Phase

The usage phase involves using the digital twin for predictive maintenance, operational optimization, and decision-making.

(a)

Predictive maintenance

The digital twin is used to predict when components or systems are likely to fail based on current operating conditions and historical data:

$${\hat{T}}_f=\mathbb{E}\left[T_f|{\mathcal{D}}_t,{\theta }^{\left(t\right)}\right]$$

(5)

where $\mathbb{E}[·]$ is the expectation operator, which in statistics refers to the expected value or mean of a random variable. In this context, it calculates the expected time to failure based on the available data and the model.

(b)

Operational optimization

Maintenance schedules are optimized to minimize downtime and cost while ensuring safety:

$$Minimize \mathcal{O}\{{\theta }^{\left(t\right)}, {\mathit{x}}_t\}$$

Real-time data are continuously monitored against the digital twin to detect anomalies $S_t$ that may indicate potential issues:

$$S_t={‖{\mathit{y}}_t-f\{{\mathit{x}}_t;{\theta }^{(t)}‖}^2$$

Anomaly detection tools identify deviations from expected behavior. Let ${\mathbf{y}}_t$ represent the actual observed outcomes and ${\hat{\mathbf{y}}}_{\mathrm{t}}$ represent the predicted outcomes from the digital twin model. The anomaly score can be computed as:

$$Anomaly Score={‖{\mathbf{y}}_t-{\hat{\mathbf{y}}}_{\mathrm{t}}‖}^2$$

(c)

Decision support

Operational recommendations are provided to decision-makers based on simulations and predictions generated by the digital twin on the base of decision rules:

$$If Anomaly Score>Threshold\to Initiate Inspection$$

Historical data and predictive models are used to inform long-term strategies such as fleet management and resource allocation on the base of strategic model:

$$Plan=\arg ma{x}_{Plan}\mathbb{E}[Operational Efficiency]$$

4.

Framework Integration

The framework for designing and supporting aircraft digital twins integrates the above phases into a cohesive system that continuously enhances the accuracy and utility of the digital twin. This integration ensures that the digital twin remains a valuable tool throughout the aircraft’s lifecycle.

(a)

Data management and integration

Robust data pipelines and integration mechanisms are established to ensure that data flow seamlessly from various sources (e.g., sensors, MRO systems) into the digital twin framework:

$$Pipeline=\{Data Sources\to Preprocessing\to Model Update\}$$

(b)

Collaboration and governance

Federated learning is implemented to enable collaboration among stakeholders without compromising data privacy or security. Governance frameworks are established to ensure that all stakeholders adhere to data sharing and model update protocols:

$$Governance=\{Data Sharing Rules,Model Aggregation Protocols\}$$

(c)

Feedback loops and continuous improvement

Feedback loops are created, where insights and outcomes from the usage phase are fed back into the design and support phases. This ensures continuous improvement in the digital twin on the base of the feedback mechanism:

$$Feedback=\{Usage Insight\to Model Refinement\}$$

This framework for designing and supporting aircraft digital twins provides a structured approach to creating an accurate, real-time virtual representation of an aircraft that evolves alongside the physical asset. By integrating model creation, continuous support, and effective usage, this framework ensures that the digital twin serves as a powerful tool for predictive maintenance, operational optimization, and decision-making throughout the aircraft’s lifecycle. The use of advanced technologies within this framework further enhances the digital twin’s capabilities, making it a critical component of modern aircraft management.

3.5. Modern Technologies as Component of ADT Framework

The second component of the ADT framework (Figure 1) is modern technologies.

As shown in previous the section, the creation and maintenance of a digital twin during an aircraft’s operational phase have historically faced significant challenges. These challenges stem from the variability in operational environments, the fragmented nature of maintenance records, and the complex ownership changes that aircraft often undergo. Fortunately, modern technologies are beginning to offer solutions to these issues, paving the way for more accurate and comprehensive digital twins.

• IoT sensors

One of the primary challenges in creating operational digital twins is the collection and integration of real-time data from various aircraft systems. Advanced IoT sensors and edge computing solutions are poised to revolutionize this aspect of digital twin technology. Next-generation IoT sensors, designed specifically for aerospace applications, can provide more comprehensive and accurate data on an aircraft’s condition and performance. These sensors, capable of withstanding extreme environmental conditions, can monitor everything from engine performance and structural integrity to cabin environmental conditions.

IoT sensors collect data from various parts of the aircraft. Let ${\mathcal{D}}_t=\{{\mathbf{x}}_t^i|i=\overline{1,N}\}$ represent the data collected at time $t$ from $N$ different sensors.

The data collected by each sensor can be modeled as:

$${\mathbf{x}}_t^i=SensorFunctio{n}_i\left({\mathbf{p}}_t\right)+{ϵ}_i$$

where

• ${\mathbf{p}}_t$ are physical parameters (e.g., temperature, pressure, vibration) affecting the sensor.
• ${ϵ}_i$ is noise or error in the sensor reading.

2.

Edge computing

Edge computing complements these advanced sensors by processing data directly on the aircraft. This approach reduces the volume of data that needs to be transmitted, addressing bandwidth limitations during flight. Edge computing systems can perform initial data analysis, identifying critical information and anomalies in real-time. This not only enables more efficient updates to the digital twin model but also allows for immediate response to potential issues, enhancing safety and operational efficiency.

Edge computing processes the raw sensor data locally on the aircraft to filter, aggregate, and preprocess the data before sending them to the central system. Let ${\mathcal{D}\prime }_t=EdgeProcess({{\mathcal{D}}^{\prime }}_t)$ represent the processed data after edge computing:

$${\mathcal{D}\prime }_t=\{{{\mathbf{x}}_t^i}^{\prime }|i=\overline{1,N}\}$$

where ${{\mathit{x}}_t^i}^{\prime }$ is processed data vector from sensor $i$, which could involve denoising, feature extraction, and normalization.

The edge processing function can be modeled as:

$${{\mathbf{x}}_t^i}^{\prime }=f_{edge}({\mathbf{x}}_t^i)$$

where $f_{edge}(·)$ is edge processing function applied to the raw data vector.

3.

Blockchain for secure maintenance records

The fragmentation of maintenance records and the loss of historical data when aircraft change ownership have been significant obstacles in creating comprehensive digital twins. Blockchain technology offers a promising solution to this challenge. By creating a decentralized, immutable ledger of all maintenance activities, modifications, and operational data, blockchain can ensure a complete and tamper-proof history of an aircraft throughout its lifecycle. This technology can facilitate seamless transfer of maintenance records between owners and operators, ensuring that no critical information is lost.

Moreover, smart contracts built on blockchain platforms can automate and standardize maintenance record-keeping processes. These contracts can trigger automatic updates to the digital twin when maintenance is performed or when specific operational milestones are reached. This level of automation and standardization can significantly reduce errors in record-keeping and ensure that digital twin models are consistently updated with accurate information.

The blockchain record at time $t$ can be modeled as:

$${\mathcal{B}}_t=H({\mathcal{B}}_{t-1}\parallel {\mathcal{M}}_t\parallel timestam{p}_t$$

where ${\mathcal{B}}_t$—the current block in the blockchain, ${\mathcal{B}}_{t-1}$—the previous block in the blockchain, ${\mathcal{M}}_t$—the maintenance data being added, $timestam{p}_t$—the time when the block is created, and $H(·)$—the cryptographic hash function.

4.

Artificial intelligence and machine learning

AI and ML technologies are set to play a crucial role in overcoming the challenges of data integration and analysis. Advanced AI algorithms can process vast amounts of heterogeneous data from various sources, including sensor readings, maintenance records, and environmental data. These systems can identify patterns, predict potential issues, and optimize performance in ways that would be impossible for human analysts.

Machine learning models, trained on historical data from entire fleets of aircraft, can become increasingly accurate in predicting wear and tear, optimizing maintenance schedules, and even suggesting design improvements for future aircraft models. These AI-driven insights can continuously refine and improve the accuracy of digital twin models, making them valuable tools for decision-making throughout the aircraft’s lifecycle.

AI and ML algorithms are used to analyze data and make predictions, such as estimating the time to failure of a component. Considering the previously introduced designations, the model is trained using

$${\theta }^{(t+1)}={\theta }^t-\eta {\nabla }_{\theta }\mathcal{L}\{{\theta }^{(t)},{\mathcal{D}}_{historical})$$

for prediction

$${\hat{T}}_f=\mathbb{E}[T_f|{{\mathcal{D}}^{\prime }}_t, {\theta }^{(t)}]$$

The challenge of standardizing data formats and integration processes across the industry can be addressed through the development of comprehensive digital platforms and industry-wide initiatives. Cloud-based platforms, designed specifically for aviation digital twins, can provide standardized interfaces for data input, analysis, and visualization. These platforms can incorporate APIs that allow seamless integration with various data sources, including aircraft systems, maintenance databases, and environmental monitoring networks.

Industry consortia and regulatory bodies can play a crucial role in establishing standards for digital twin data formats, communication protocols, and security measures. By creating a common language for digital twins in aviation, these standards can facilitate interoperability between different systems and stakeholders, from manufacturers and operators to maintenance providers and regulatory agencies.

5.

Augmented reality and virtual reality

Augmented reality (AR) and virtual reality (VR) technologies offer innovative solutions for bridging the gap between physical aircraft and their digital representations. AR systems can overlay digital twin data onto physical aircraft during maintenance procedures, allowing technicians to visualize historical data, predict potential issues, and access real-time guidance. This can significantly enhance the accuracy and efficiency of maintenance operations while ensuring that the digital twin remains synchronized with the physical aircraft.

VR environments, on the other hand, can provide immersive interfaces for interacting with digital twin data. Engineers and analysts can “walk through” virtual representations of aircraft, examining complex systems and simulating various operational scenarios. This level of interaction can lead to deeper insights and more effective decision-making throughout the aircraft’s lifecycle.

AR and VR are used to visualize the digital twin data in real-time:

$${\mathcal{V}}_{AR}=AR\left({\mathcal{M}}_t,{\mathit{x}}_{real}\right), {\mathcal{V}}_{VR}=VR\left({\mathcal{M}}_t\right)$$

where ${\mathcal{V}}_{AR}$ and ${\mathcal{V}}_{VR}$—the augmented reality and virtual reality visualization.

6.

The role of 5G and 6G networks

Current 5G and future 6G networks will play a crucial role in enabling real-time data transmission between aircraft and ground systems. These high-speed, low-latency networks can facilitate the continuous update of digital twin models, even for aircraft in flight. This constant connectivity ensures that digital twins remain accurate representations of the current state of the aircraft, enabling more responsive and proactive management of operations and maintenance.

The communication infrastructure for real-time data transmission between the aircraft and ground systems is provided by 5G and 6G networks. The real-time update of the digital twin model can be expressed as:

$${\mathcal{M}}_{t+1}={\mathcal{M}}_t+∆{\mathcal{M}}_t$$

where $∆{\mathcal{M}}_t$ is the model update based on the latest data transmitted over the network.

7.

Quantum computing

As quantum computing technology matures, it has the potential to revolutionize the complexity and scale of simulations possible within digital twin models. Quantum computers could process vast amounts of data and perform complex calculations that are currently infeasible, allowing for more accurate modeling of aircraft performance under a wide range of conditions and scenarios.

Quantum computing can be used to perform complex simulations and optimizations that are currently infeasible with classical computing:

$${\mathcal{Q}}_{sim}=QuantumSim({\mathcal{M}}_t,{\mathit{X}}_{quantum})$$

where ${\mathcal{Q}}_{sim}$ is result of the quantum simulation and ${\mathit{X}}_{quantum}$ is the set of quantum inputs (e.g., qubits representing different conditions).

The integration of these technologies—IoT, edge computing, blockchain, AI/ML, AR/VR, advanced networking, and potentially quantum computing—creates a powerful ecosystem for overcoming the challenges of creating and maintaining digital twins for operational aircraft. This technological convergence promises to transform the fragmented, often incomplete picture of an aircraft’s status into a comprehensive, real-time, and predictive digital representation.

Implementing these solutions will require significant investment and collaboration across the aviation industry. However, the potential benefits are immense. Accurate digital twins can lead to enhanced safety, improved operational efficiency, reduced maintenance costs, and more sustainable aircraft operations. They can inform better design decisions for future aircraft, optimize fleet management, and even revolutionize air traffic control systems.

3.6. Models as Component of ADT Framework

The third component of the ADT framework (Figure 1) is models for digital twins. The mathematical models used to build digital twins for aircraft lifecycle management can be categorized into three primary types as follows: physics-based models, data-driven models, and hybrid models. Each type uses different mathematical techniques to simulate, predict, and optimize the behavior and condition of the aircraft throughout its lifecycle.

• Physics-based models

Physics-based models rely on the principles of physics and engineering to simulate the behavior of an aircraft’s systems under various conditions. These models typically involve differential equations, finite element analysis, and computational fluid dynamics.

(a)

Structural dynamics model

The structural dynamics of an aircraft can be modeled using the finite element method. This involves solving the following system of equations:

$$\mathit{F}\left(\mathrm{t}\right)=\mathit{M}\ddot{\mathit{u}}\left(t\right)+\mathit{P}\dot{\mathit{u}}\left(t\right)+\mathit{{\rm K}}\mathit{u}(t)$$

where $\mathit{F}\left(\mathrm{t}\right)$ is the external force vector at time $t$, $\mathit{M}$ is the mass matrix, $\mathbf{P}$ is the damping matrix, $\mathit{{\rm K}}$ is the stiffness matrix, $\mathit{u}(t)$ is the displacement vector at time $t$, and $\ddot{\mathbf{u}}\left(t\right)$ and $\dot{\mathit{u}}\left(t\right)$ represent acceleration and velocity vectors, respectively.

This equation models the dynamic response of the aircraft’s structure to various forces, such as aerodynamic loads, during flight.

(b)

Aerodynamic model

Aerodynamic behavior is often modeled using Navier–Stokes equations, which describe the motion of fluid substances like air:

$${\displaystyle \frac{\partial \mathbf{u}}{\partial t}}+\left(\mathbf{u}·\nabla \right)\mathbf{u}=-{\displaystyle \frac{1}{\rho }}\nabla p+\upsilon {\nabla }^2\mathbf{u}+\mathbf{f}$$

where $\mathbf{u}$ is the velocity field, $p$ is the pressure field, $\rho$ is the fluid density, $\upsilon$ is the kinematic viscosity, and $\mathbf{f}$ represents body forces (e.g., gravity).

These partial differential equations are solved numerically using methods such as computational fluid dynamics to predict the airflow around the aircraft and its aerodynamic forces.

(c)

System behaviors

System behaviors, such as those of the propulsion system or avionics, can be modeled using differential equations that represent the dynamics of these systems. For example, the behavior of an aircraft’s engine might be modeled by:

$${\displaystyle \frac{d\mathbf{x}}{dt}}=\mathbf{A}\mathbf{x}\left(t\right)+\mathbf{B}\mathbf{u}(t)$$

where $\mathbf{x}\left(t\right)$ is the state vector representing engine parameters (e.g., temperature, pressure) at time $t$, $\mathbf{A}$ is the system matrix that defines how the state variables evolve over time, $\mathbf{u}(t)$ is the input vector (e.g., throttle position), and $\mathbf{B}$ is the input matrix that maps the inputs to the state variables.

2.

Data-driven models.

Data-driven models use statistical and machine learning techniques to predict and analyze the aircraft’s performance and condition based on historical data and real-time sensor inputs.

(a)

Regression models

A common data-driven approach is regression analysis, which models the relationship between a dependent variable (e.g., component failure time) and one or more independent variables (e.g., operating conditions):

$$y={\beta }_0+\sum _{i=1}^n{\beta }_i{x}_i+ϵ$$

where $y$ is the dependent variable, $x_i$ are the independent variables, ${\beta }_i$ are the coefficients, and $ϵ$ represents the error term.

Regression models can be linear or nonlinear, depending on the nature of the relationship between the variables.

(b)

Machine learning models

Machine learning models, such as neural networks, are used for more complex predictions. A neural network model with $L$ layers can be represented as:

$${\mathbf{a}}^{(l+1)}=\sigma [{\mathbf{W}}^{\left(l\right)}{\mathbf{a}}^{\left(l\right)}+{\mathbf{b}}^{\left(l\right)}]$$

where ${\mathbf{a}}^{\left(l\right)}$ s the activation vector of the $l$-th layer, ${\mathbf{W}}^{\left(l\right)}$ is the weight matrix for the $l$-th layer, ${\mathbf{b}}^{\left(l\right)}$ is the bias vector for the $l$-th layer, and $\sigma (·)$ is the activation function.

Machine learning models can learn from large datasets to predict outcomes like remaining useful life of components, potential failures, and optimal maintenance schedules.

(c)

Time series analysis

To predict future events based on past data, time series analysis is often used. A basic time series model could be an autoregressive model:

$$y_t=c+\sum _{i+1}^m{\varphi }_i{y}_{t-1}+{ϵ}_t$$

where $y_t$ is the value of the time series at time $t$, $c$ is a constant, ${\varphi }_i$ are the parameters of the model, $m$ is the number of lagged observations (the order of the model), and ${ϵ}_t$ is the error term at time $t$.

3.

Hybrid models

Hybrid models combine physics-based and data-driven approaches to leverage the strengths of both methodologies. These models typically involve a combination of numerical simulation and machine learning.

(a)

Kalman filter

The Kalman filter is a recursive algorithm used to estimate the state of a dynamic system by combining a physical model (prediction) with noisy measurements (observations):

Prediction:

$${\mathbf{x}}_{k|k-1}={\mathbf{A}}_{k-1}{\mathbf{x}}_{k-1|k-1}+{\mathbf{B}}_{k-1}{\mathbf{u}}_{k-1}$$

$${\mathbf{P}}_{k|k-1}={\mathbf{A}}_{k-1}{\mathbf{P}}_{k-1|k-1}{\mathrm{A}}_{k-1}^T+{\mathbf{Q}}_{k-1}$$

Update:

$${\mathbf{K}}_k={\mathbf{P}}_{k|k-1}{\mathbf{H}}_k^T{({\mathbf{H}}_k{\mathbf{P}}_{k|k-1}{\mathbf{H}}_k^T+{\mathbf{R}}_k)}^{-1}$$

$${\mathbf{x}}_{k|k}={\mathbf{x}}_{k|k-1}+{\mathbf{K}}_k({\mathbf{z}}_k-{\mathbf{H}}_k{\mathbf{x}}_{k|k-1})$$

$${\mathbf{P}}_{k|k}=(\mathbf{I}-{\mathbf{K}}_k{\mathbf{H}}_k){\mathbf{P}}_{k|k-1}$$

where ${\mathit{x}}_{k|k-1}$ is the predicted state vector at time $k$, ${\mathit{P}}_{k|k-1}$ is the predicted state covariance matrix, ${\mathit{K}}_k$ is the Kalman gain, ${\mathit{z}}_k$ is the measurement vector, and ${\mathbf{A}}_k, {\mathbf{B}}_k, {\mathbf{H}}_k, {\mathbf{Q}}_k, \mathrm{a}\mathrm{n}\mathrm{d} {\mathbf{R}}_k$ are matrices related to system dynamics and noise covariances.

Kalman filters are useful to estimate the current state of an aircraft (e.g., position, velocity) by blending physical models with real-time sensor data, making them essential in hybrid digital twins.

(b)

Surrogate models

Surrogate models are simplified models that approximate more complex physics-based simulations using data-driven methods. For instance, a surrogate model might be a polynomial approximation of a computational fluid dynamics simulation:

$$\hat{f}\left(\mathbf{x}\right)=\sum _{i=0}^n{\alpha }_i{\mathbf{x}}^i$$

where $\hat{f}\left(\mathit{x}\right)$ is the surrogate model approximation, $\mathbf{x}$ is the input vector, and ${\alpha }_i$ are the coefficients determined through fitting.

These models reduce computational costs while retaining sufficient accuracy for certain applications, such as optimization and real-time decision-making.

Mathematical models form the backbone of digital twins in aircraft lifecycle management. Physics-based models simulate the physical behavior of the aircraft using differential equations and numerical methods. Data-driven models utilize statistical and machine learning techniques to predict outcomes based on historical data. Hybrid models combine these approaches to create more robust and dynamic digital twins, capable of real-time updates and predictions. Together, these models ensure that the digital twin accurately reflects the physical aircraft throughout its lifecycle, enabling better decision-making and optimizing performance, maintenance, and safety.

3.7. Supporting Components in the ADT Framework

The fourth component of the digital twin framework is supporting components. They ensure the seamless integration and operation of the lifecycle phases, technologies, and models. This component is essential for maintaining the accuracy, reliability, and effectiveness of the digital twin.

• Data management and integration

Data management and integration involve the seamless flow of data from various sources into the digital twin, ensuring that the twin remains up-to-date and accurate. This process can be mathematically described as follows:

(a)

Data collection and aggregation

Let ${\mathcal{D}}_t^i$ represent the data collected from source $i$ at time $t$, where $i=\overline{1,N}$ and $N$ is the number of data sources (e.g., sensors, databases, logs). The aggregated dataset at time $t$ can be expressed as:

$${\mathcal{D}}_t=\bigcup _{i=1}^N{\mathcal{D}}_t^i$$

The aggregated data ${\mathcal{D}}_t$ includes information from all sources, providing a comprehensive view of the aircraft’s state at time $t$.

(b)

Data preprocessing and transformation

Before the data are integrated into the digital twin, preprocessing is performed to remove noise, handle missing values, and normalize the data. Let ${\mathcal{D}\prime }_t$ represent the preprocessed data:

$${\mathcal{D}\prime }_t=Preprocess({\mathcal{D}}_t)$$

where $Preprocess(·)$ is a function that applies various preprocessing steps, such as filtering, imputation, and normalization, to the raw data.

(c)

Data integration

The preprocessed data ${\mathcal{D}\prime }_t$ are then integrated into the digital twin model. The integration process can be described as:

$${\mathcal{M}}_t={\mathcal{M}}_{t-1}+Integrate({{\mathcal{D}}^{\prime }}_t)$$

where ${\mathcal{M}}_t$ is the updated digital twin model at time $t$ and $Integrate(·)$ is a function that incorporates the new data into the existing model, ensuring that the digital twin accurately reflects the current state of the aircraft.

2.

Federated learning and collaboration tools.

Federated learning and collaboration tools enable multiple stakeholders to collaboratively train and update the digital twin without sharing raw data. This approach maintains data privacy while using the collective knowledge of all stakeholders.

Local model training, model update aggregation and distribution of the updated model are defined by expressions (1)–(4).

3.

Analytics and decision support tools

Analytics and decision support tools provide actionable insights that aid in predictive maintenance scheduling, operational optimization, and anomaly detection.

(a)

Predictive analytics for maintenance

The predictive maintenance model uses historical and real-time data to estimate the time to failure $T_f$ for critical components. This prediction can be expressed as (5):

$${\hat{T}}_f=\mathbb{E}[T_f|{\mathcal{D}}_t,{\theta }^{\left(t\right)}]$$

(b)

Operational optimization

Operational optimization involves finding the optimal operational parameters ${\mathbf{x}}_t$ that minimize or maximize a specific objective function $\mathcal{O}$, such as fuel consumption or flight time:

$${\mathbf{x}}_t^{*}=\arg \underset{x_i}{\min }\mathcal{O}[{\mathbf{x}}_t,{\mathcal{D}}_t, {\theta }^{\left(t\right)}]$$

where ${\mathbf{x}}_t^{*}$ are the optimal operational parameters and $\mathcal{O}(·)$ is the objective function to be optimized.

(c)

Anomaly detection

Anomaly detection tools identify deviations from expected behavior. Let ${\mathbf{y}}_t$ represent the actual observed outcomes and ${\hat{\mathbf{y}}}_{\mathrm{t}}$ represent the predicted outcomes from the digital twin model. The anomaly score $S_t$ can be computed as:

$$S_t={‖{\mathbf{y}}_t-{\hat{\mathbf{y}}}_{\mathrm{t}}‖}^2$$

where ${‖·‖}^2$ denotes the squared Euclidean norm, representing the difference between observed and predicted outcomes. If $S_t$ exceeds a predefined threshold $\tau$, an anomaly is detected.

The supporting components of the digital twin framework—data management and integration, federated learning and collaboration tools, and analytics and decision support tools—are mathematically described through processes that ensure seamless data flow, collaborative model training, and actionable insights. Data management and integration ensure that the digital twin remains up-to-date and accurate, federated learning enables secure collaboration among stakeholders, and analytics tools provide critical decision-making support. Together, these components form the backbone of a robust and effective digital twin ecosystem, facilitating continuous improvement and optimization throughout the aircraft’s lifecycle.

4. Discussion

4.1. Differentiating Digital Twins from Real-Time Monitoring Systems

A digital twin is a dynamic, real-time virtual representation of a physical system or asset that not only reflects the current state of the system but also provides predictive analytics, decision support, and the ability to simulate future scenarios. The key features that distinguish a digital twin from more straightforward real-time monitoring systems are as follows:

• Digital twins use data-driven and physics-based models to predict future behaviors and potential failures. This predictive capability allows for proactive maintenance and optimization rather than reactive responses.
• A digital twin can simulate various operational scenarios and predict the outcomes of different decisions. This goes beyond simply monitoring the system in real time and allows stakeholders to test different approaches in a virtual environment.
• Digital twins provide insights that support strategic decision-making. They are not only concerned with monitoring but also assist in optimizing operations, improving efficiency, and reducing downtime through integrated analytics.

In contrast, real-time monitoring systems primarily focus on observing the current state of a system, collecting and reporting data, and potentially raising alarms if certain thresholds are exceeded. These systems typically lack the predictive and simulation capabilities of digital twins and are more reactive in nature. They are valuable for short-term operational awareness but do not provide the long-term forecasting or decision support that digital twins offer.

While digital twins share some similarities with real-time monitoring systems, they represent a significant advancement in terms of capabilities, scope, and impact on asset management. It is crucial to understand these differences to appreciate the full transformative potential of digital twins in aircraft lifecycle management.

• Predictive Capabilities

Real-time monitoring systems primarily focus on collecting and displaying current data, often with the ability to show historical trends. Their primary function is to alert operators to immediate issues or deviations from normal parameters. In contrast, digital twins leverage these real-time and historical data as a foundation for predictive analytics. By employing advanced machine learning algorithms and physics-based models, digital twins can forecast the future states of the aircraft, predict potential failures, and anticipate maintenance needs. This predictive capability allows for proactive decision-making, optimizing maintenance schedules, and preventing unplanned downtime.

2.

Bi-Directional Data Flow

Traditional monitoring systems typically have a unidirectional flow of information: data are collected from the physical asset and presented to human operators or automated systems. Digital twins, however, facilitate a bi-directional flow of information. Not only do they receive data from the physical aircraft, but they can also send information back, influencing the operation and management of the physical asset. For example, insights gained from simulations run on the digital twin can be used to adjust the operational parameters of the actual aircraft, creating a continuous feedback loop for optimization.

3.

Simulation and Scenario Testing

Unlike simple monitoring tools, digital twins have the capability to run complex simulations and scenario tests. This feature allows engineers and operators to explore “what-if” scenarios without risking the physical asset. For instance, the impact of different flight paths on fuel consumption, the effects of various environmental conditions on engine performance, or the consequences of different maintenance schedules can all be simulated and analyzed. This capability is particularly valuable in aviation, where physical testing can be extremely costly and time-consuming.

4.

Holistic System Representation

While monitoring systems often focus on specific components or subsystems, digital twins aim to provide a comprehensive representation of the entire aircraft. This holistic approach allows for a deeper understanding of the interactions between different systems and how these interactions affect overall performance. For example, a digital twin can model how changes in the aircraft’s aerodynamics might impact engine performance, fuel efficiency, and structural load—interconnections that might be missed by isolated monitoring systems.

5.

Lifecycle Management

Digital twins are designed to evolve alongside the physical asset throughout its entire lifecycle. From the design phase through manufacturing, operation, and eventual decommissioning, the digital twin accumulates data and refines its models. This longitudinal perspective allows for continuous improvement and optimization over the aircraft’s lifespan. In contrast, traditional monitoring systems typically focus on the operational phase, with limited integration of design or end-of-life considerations.

6.

Sophisticated Decision Support

While monitoring systems can trigger alerts based on predefined thresholds, digital twins offer much more sophisticated decision-support capabilities. By combining real-time data, historical trends, predictive analytics, and simulation results, digital twins can provide nuanced recommendations for optimal courses of action. These recommendations can range from immediate operational adjustments to long-term strategic decisions about fleet management or aircraft design modifications.

7.

Knowledge Accumulation and Transfer

Digital twins serve as a centralized repository of knowledge about the aircraft. They accumulate insights from the design, manufacturing, operation, and maintenance phases, creating a rich source of information that can inform future designs and improve overall system performance. This knowledge accumulation and transfer capability extends beyond a single aircraft, potentially benefiting entire fleets or even future aircraft designs.

8.

Integration with Emerging Technologies

Digital twins are well-positioned to integrate with and leverage emerging technologies such as artificial intelligence, edge computing, and advanced IoT sensors. While monitoring systems can also benefit from these technologies, digital twins provide a more comprehensive framework for their integration, allowing for more sophisticated analysis and decision-making capabilities.

While real-time monitoring systems play a crucial role in aircraft management, digital twins represent a significant leap forward in terms of capabilities and potential impact. By providing a dynamic, holistic, and predictive virtual representation of the physical asset, digital twins offer new opportunities for optimization, risk reduction, and value creation throughout the aircraft lifecycle.

4.2. Advantages of Hybrid Models over Standalone Data-Driven or Physics-Based Models

The integration of physics-based and data-driven models into hybrid models offers significant advantages over using either approach in isolation. Each type of model has its strengths and limitations, and the hybrid approach aims to use the best of both model types, resulting in more accurate, robust, and scalable systems. Below, we outline the key advantages of hybrid models in the context of digital twin technology for aircraft lifecycle management.

• Improved Predictive Accuracy

One of the most critical advantages of hybrid models is the improvement in predictive accuracy. Physics-based models are built on well-established engineering principles and are highly accurate in simulating known physical phenomena. However, they can struggle with predicting complex behaviors that arise from unknown factors, high variability, or nonlinear processes that are difficult to model mathematically. Conversely, data-driven models excel in pattern recognition and prediction based on large datasets, but they may suffer from overfitting or inaccuracies when faced with scenarios not well-represented in the data.

By combining the two, hybrid models benefit from the high accuracy and generalization capabilities of physics-based models while using data-driven models to handle unknown, complex, or emergent behaviors. This synergy ensures that the hybrid model can accurately predict both the expected performance of an aircraft system and the potential impact of unforeseen conditions.

2.

Adaptability and Real-Time Updating

Hybrid models offer a unique advantage in terms of adaptability. Physics-based models are often static, relying on predetermined equations and principles that do not change as the system evolves. On the other hand, data-driven models are inherently dynamic, as they continuously update based on new data inputs.

In a hybrid model, the physics-based component provides a stable foundation, while the data-driven component allows the system to adapt to real-time conditions. For instance, during the operation phase of an aircraft, sensor data collected in real time can be fed into the data-driven aspect of the hybrid model. This enables the model to adjust its predictions and simulations dynamically, even when faced with evolving operational conditions or unexpected events, ensuring that the digital twin remains a highly accurate reflection of the physical aircraft.

3.

Handling Sparse or Uncertain Data

One common challenge in aircraft lifecycle management is the availability of sparse or uncertain data, especially during the early stages of a system’s operation or when dealing with new designs. Physics-based models excel in these scenarios because they can predict system behavior based on theoretical principles without needing large amounts of operational data. However, these models may not capture the full complexity of the system, especially in situations where operational factors differ from theoretical assumptions.

By incorporating data-driven elements, hybrid models can compensate for the incomplete or uncertain data that physics-based models struggle with. Data-driven models, once trained on small but meaningful datasets, can fill in the gaps where empirical data are lacking or incomplete. This capability makes hybrid models more resilient to data limitations while providing more robust predictions.

4.

Reduction of Computational Complexity

Physics-based models often require significant computational resources, especially for complex systems such as aircraft, where detailed simulations of structural dynamics, aerodynamics, and material properties are involved. These models can be computationally expensive and time-consuming to run, limiting their utility in real-time applications or scenarios where quick decision-making is necessary.

Data-driven models, by contrast, are typically faster to execute once trained, making them well-suited for real-time applications. In hybrid models, data-driven components can be used to approximate certain aspects of the system’s behavior, significantly reducing the computational complexity required for full-scale physics-based simulations. This allows for faster, real-time simulations and predictions without compromising on accuracy. As a result, hybrid models can balance accuracy and efficiency, enabling real-time decision-making in complex operational environments.

5.

Enhanced Fault Detection and Diagnosis

The integration of physics-based and data-driven models improves the ability to detect and diagnose faults in aircraft systems. Physics-based models can simulate how a system should behave under normal conditions, providing a benchmark for expected performance. When discrepancies arise between the predicted and actual system behavior, the data-driven component can analyze the deviation patterns to detect early signs of failure or degradation.

For example, in predictive maintenance, the hybrid model can identify subtle changes in system behavior that might not be noticeable through physics-based predictions alone. The data-driven model can recognize trends and correlations in sensor data, enabling earlier and more accurate detection of potential issues. This combination of model-based reasoning and data-driven insights enhances the overall reliability and safety of the aircraft, reducing the likelihood of unexpected failures.

6.

Better Generalization Across Different Use Cases

Another significant advantage of hybrid models is their ability to generalize across different use cases. Physics-based models are often tailored to specific system configurations and operating conditions, which may limit their flexibility. Data-driven models, on the other hand, are more adaptable to variations in the system and can be retrained or fine-tuned for new scenarios.

Hybrid models provide the flexibility needed to handle a broader range of conditions, including new aircraft designs, varying operational environments, and different maintenance practices. The physics-based core ensures that the model adheres to fundamental principles, while the data-driven component allows it to adapt to new data as they becomes available. This makes hybrid models more versatile and applicable to a wide range of use cases throughout the aircraft lifecycle.

4.3. Decision-Making in Aviation Digital Twin Framework

The proposed framework for the formation of an aviation digital twin is not an end in itself; rather, it is a crucial tool for facilitating decision-making throughout the operation of aviation equipment. The primary purpose of this framework is to enable informed, data-driven decisions that enhance the safety, efficiency, and performance of aircraft over their entire lifecycle. Therefore, in addition to understanding the components of the digital twin framework, it is essential to consider the main models of decision-making that come into play when using digital twins.

Digital twins provide a virtual replica of physical aircraft systems, continuously updated with real-time data and capable of running simulations to predict future behavior. This capability allows stakeholders, from engineers to pilots and maintenance crews, to make well-informed decisions based on accurate, up-to-date information. However, the effectiveness of these decisions depends on the integration of robust decision-making models within the digital twin framework.

The following key decision-making models are integral to the digital twin framework:

• The predictive maintenance decision model aims to decide when and how maintenance should be performed to prevent failures and extend the lifespan of components. This model uses real-time and historical data to predict the remaining useful life of aircraft components, incorporating stochastic models such as the Weibull distribution and Bayesian updating to refine predictions. Based on these estimates, the model recommends optimal maintenance schedules, balancing the risks of failure against the costs of maintenance, enabling proactive maintenance, and reducing unplanned downtime.
• The operational optimization model focuses on optimizing flight operations for fuel efficiency, safety, and performance. Using real-time operational data, this model continuously adjusts flight parameters such as speed, altitude, and engine settings through optimization algorithms. The goal is to minimize fuel consumption while ensuring safety and compliance with operational constraints. The model provides real-time recommendations to pilots or autonomous systems, facilitating dynamic adjustments in response to changing environmental factors.
• The design iteration decision model integrates feedback loops from various lifecycle phases, including operation, maintenance, and production, to improve aircraft design based on operational feedback and evolving technological requirements. This model employs regression analysis and machine learning to identify design elements that need improvement and to predict the impact of potential design changes. Engineers can then use these insights to iterate on the design, ensuring continuous improvement over successive generations of the aircraft.
• In the case of emergencies or anomalies during flight, the incident response decision model uses real-time data and scenario simulations to evaluate different response strategies. Incorporating decision trees and risk assessment algorithms, this model determines the best course of action, providing immediate recommendations to the flight crew or autonomous systems to minimize risk and ensure safety.
• The lifecycle cost management model is designed to optimize the total cost of ownership of the aircraft throughout its lifecycle. By integrating data from all lifecycle phases, this model uses cost modeling and optimization techniques to forecast and minimize lifecycle costs, considering factors such as fuel consumption, maintenance expenses, and depreciation. Strategic recommendations are then provided to management, aiding in investment decisions, budget planning, and the scheduling of upgrades or replacements based on cost-benefit analysis.

To realize the full benefits of the digital twin framework, these decision-making models must be seamlessly integrated into the digital twin’s architecture. This integration ensures that the digital twin not only provides a virtual representation of the aircraft but also serves as a decision-support tool that continuously guides operational, maintenance, and design decisions.

4.4. Knowledge-Driven Framework of Aircraft Digital Twins

The evolution of aircraft digital twins has significantly advanced with the integration of knowledge-driven frameworks, which enhance the twin’s ability to simulate, predict, and optimize aircraft operations. This approach uses vast amounts of data and knowledge from various sources, transforming them into actionable insights that support decision-making throughout the aircraft’s lifecycle. Figure 3 provided a visual representation of this knowledge-driven framework, highlighting its key components and their interconnections.

The framework begins with the collection of knowledge from the following two primary areas: knowledge about external environment behavior and knowledge about system behavior. These knowledge sources are crucial for understanding both the operational context in which the aircraft functions and the internal dynamics of the aircraft itself. Knowledge about the external environment includes data and information related to the conditions that affect aircraft performance, such as weather, air traffic, and geographical factors. Sources of this knowledge include sensors, which collect real-time environmental data, documentation, such as historical records and environmental reports, and user experience, such as insights from pilots and operators who interact with the aircraft in varying conditions. On the other hand, knowledge about system behavior encompasses the internal workings of the aircraft, including its structural integrity, system performance, and operational efficiency. This knowledge is gathered from the aircraft’s structure, design and engineering specifications, sensors that monitor the health of various aircraft systems in real-time, documentation like maintenance records and technical manuals, and user experience, including feedback from maintenance crews, engineers, and other stakeholders.

Once the knowledge is collected, it is processed and organized into several databases and knowledgebases. Each knowledge processor is tailored to handle specific types of data and information, ensuring that the knowledge is accurately interpreted and stored in the most relevant database. The system database stores information related to the aircraft’s systems, capturing data on system performance, operational parameters, and maintenance activities. The user database contains user-generated knowledge, including insights from pilots, maintenance personnel, and other users who interact with the aircraft. The personal knowledgebase holds knowledge that is specific to individual aircraft or specific user experiences, ensuring that personalized data are accessible for decision-making. The common knowledgebase aggregates general knowledge that is applicable across multiple aircraft models or operational contexts, providing a broad base of information that supports overall system understanding. The domain ontology organizes knowledge according to specific domains, such as aerodynamics, avionics, or structural engineering, ensuring that the knowledge is categorized and structured in a way that aligns with domain-specific standards and terminologies. Similarly, the application ontology focuses on the specific applications of knowledge, categorizing it according to how it will be used in the digital twin’s functions, such as simulation, prediction, or optimization.

At the heart of the framework is the orchestration layer, which coordinates the flow of knowledge between the databases and the digital twin’s operational processes. The orchestration layer ensures that the right knowledge is delivered to the right place at the right time, enabling the digital twin to function effectively and respond dynamically to new data or changing conditions. Beneath the orchestration layer is the semantic repository, which plays a crucial role in managing the relationships between different pieces of knowledge. The semantic repository uses ontology-based structures to map out the connections between various data points and knowledge elements, ensuring that the digital twin has a deep, contextual understanding of the information it processes. This repository acts as the brain of the digital twin, providing a comprehensive and interconnected view of the aircraft’s knowledge landscape.

The ontology meta database underpins the entire framework. It serves as the foundational structure that defines and organizes the ontologies used within the semantic repository. By establishing the relationships and hierarchies between different types of knowledge, the ontology meta database ensures that the digital twin’s knowledge-driven processes are coherent, standardized, and aligned with industry best practices.

The knowledge-driven framework for aircraft digital twins represents a sophisticated approach to managing and leveraging vast amounts of data and expertise. By systematically collecting, processing, and orchestrating knowledge from diverse sources, this framework enables the digital twin to perform complex simulations, make accurate predictions, and support critical decision-making processes. As illustrated in the figure, the interplay between knowledge sources, databases, ontologies, and the semantic repository creates a robust and dynamic system that enhances the capabilities of the aircraft digital twin, ultimately leading to more efficient, safe, and optimized aircraft operations.

4.5. Challenges and Limitations in Implementing Aircraft Digital Twins

While the integration of knowledge-driven frameworks into aircraft digital twins represents a significant advancement in the field, there are several limitations to the study and implementation of such systems that need to be addressed. These limitations arise from the complexities involved in data management, the challenges of integrating diverse data sources, the limitations of current technologies, and the potential issues related to system scalability and security.

One of the primary limitations of the discussed study is the challenge of data quality and management. Aircraft digital twins rely on vast amounts of data from various sources, including sensors, user experience, and documentation. However, the accuracy and reliability of the digital twin are directly dependent on the quality of the data it processes. In many cases, data collected from sensors may be noisy, incomplete, or subject to various forms of bias. Additionally, data from different sources may be stored in incompatible formats or lack standardization, making integration difficult. These issues can lead to inconsistencies in the digital twin, which may result in inaccurate simulations, predictions, and decision-making processes.

Another significant limitation is the integration of diverse knowledge sources. The knowledge-driven framework relies on the ability to integrate information from various domains, such as aerodynamics, avionics, and structural engineering, as well as from different types of knowledge bases, including user databases, personal knowledgebases, and domain ontologies. However, combining these diverse sources of knowledge into a coherent and unified framework is a complex task. Differences in data formats, terminologies, and domain-specific standards can create barriers to effective integration. Moreover, ensuring that the knowledge is up-to-date and reflective of the latest developments in each domain requires continuous monitoring and updating, which can be resource-intensive and challenging to manage.

The limitations of current technologies also pose a challenge to the implementation of knowledge-driven aircraft digital twins. While technologies such as artificial intelligence, machine learning, and semantic repositories offer powerful tools for processing and analyzing data, they are not without their limitations. AI and ML models, for instance, require large datasets for training and may struggle with the inherent variability and uncertainty present in real-world aviation data. Additionally, the complexity of these models can make them difficult to interpret, leading to potential issues with transparency and explainability in decision-making processes. Similarly, while semantic repositories provide a structured way to organize and retrieve knowledge, they may struggle with the dynamic and evolving nature of the knowledge required for aircraft operations.

Scalability is another limitation that must be considered. As aircraft digital twins become more sophisticated and incorporate more data and knowledge sources, the computational resources required to maintain and operate these systems increase significantly. This can lead to challenges in scaling the system to handle large fleets of aircraft or to integrate real-time data streams from multiple sources. The need for high-performance computing infrastructure, along with the associated costs, can be a barrier to widespread adoption of digital twin technology in the aviation industry.

Security and privacy concerns are also critical limitations in the study and implementation of aircraft digital twins. The integration of data from various sources, especially sensitive data related to aircraft operations and maintenance, raises concerns about data security and privacy. Federated learning and collaboration tools, while designed to address some of these concerns, are not foolproof and may still be vulnerable to attacks or breaches. Ensuring the security and privacy of the data used in digital twins is essential to maintaining the trust and confidence of stakeholders, but achieving this goal requires ongoing vigilance and investment in cybersecurity measures.

Integrating blockchain into real-world aircraft lifecycle management poses several challenges, particularly regarding network requirements and stakeholder cooperation. Blockchain technology, especially in decentralized forms, often faces latency issues that can hinder its ability to process real-time data, which is critical in aviation operations where timely decisions are essential. While private or permissioned blockchains may reduce latency, they can also compromise some of the decentralization and transparency advantages, raising concerns about trust among stakeholders. Additionally, achieving full stakeholder cooperation is challenging, as it requires consensus from diverse entities, including manufacturers, airlines, maintenance providers, and regulators, all of whom may have conflicting interests regarding data transparency and sharing. Resistance may arise because of concerns over the costs of implementing blockchain, data privacy, or the potential exposure of proprietary or sensitive information. Therefore, addressing these technical and cooperative obstacles is crucial for blockchain’s successful integration into aircraft lifecycle management.

Finally, there are ethical and regulatory challenges associated with the deployment of knowledge-driven aircraft digital twins. As these systems become more autonomous and play a greater role in decision-making processes, questions arise about accountability and responsibility in the event of errors or failures. Regulatory frameworks for the use of digital twins in aviation are still evolving, and there may be gaps or ambiguities in the rules governing their use. Ensuring that digital twin technology is deployed in a manner that is ethical, transparent, and compliant with relevant regulations is a significant challenge that requires careful consideration.

While the knowledge-driven framework for aircraft digital twins offers promising advancements in the field of aviation, it is not without its limitations. Challenges related to data quality, integration, technology, scalability, security, and ethics must be carefully addressed to realize the full potential of this technology. By acknowledging and addressing these limitations, researchers and practitioners can work towards more robust, reliable, and effective digital twin systems that enhance the safety, efficiency, and sustainability of aircraft operations.

4.6. Future Directions of Research

As the concept of aviation digital twins continues to evolve, it opens numerous avenues for future research that could significantly enhance the technology’s capabilities and broaden its application across the aviation industry. While the current implementations of digital twins have already demonstrated their potential in improving aircraft operations, maintenance, and safety, there are still several areas where further research is needed to address existing challenges and unlock new possibilities.

One of the key areas for future research is the improvement in data integration and management within the digital twin framework. As the volume and variety of data sources continue to grow, developing more sophisticated methods for integrating heterogeneous data into a cohesive model will be crucial. This includes research into advanced data fusion techniques that can seamlessly combine real-time sensor data, historical records, and external environmental information. Additionally, research into developing more robust data preprocessing algorithms that can handle noisy, incomplete, or biased data will be essential in ensuring the accuracy and reliability of the digital twin. Another promising area is the exploration of decentralized data management approaches, such as distributed ledger technologies, which could enable secure and efficient data sharing across different stakeholders while maintaining data integrity and privacy.

Advancing predictive analytics is another critical direction for future research. The ability to predict potential failures, maintenance needs, and operational issues accurately is a cornerstone of the digital twin’s value proposition. Future research should focus on developing more sophisticated predictive models that can better capture the complex, non-linear relationships between various aircraft systems and their operating environments. This includes the use of advanced machine learning techniques, such as deep learning and reinforcement learning, which can learn from vast amounts of data and improve their predictions over time. Moreover, research into integrating uncertainty quantification into predictive models could provide more reliable and confidence-based predictions, which are crucial for decision-making in safety-critical systems like aviation.

AI and ML will continue to be at the forefront of digital twin research. However, there is a growing need to improve the interpretability and explainability of these models, especially in the context of aviation, where transparency is essential for safety and regulatory compliance. Future research could explore the development of hybrid models that combine the interpretability of physics-based models with the flexibility and accuracy of data-driven approaches. Additionally, research into federated learning could enable collaborative model training across multiple stakeholders without compromising data privacy, which is particularly important in the highly regulated aviation industry.

Cybersecurity will be an increasingly important area of research as digital twins become more integrated into the operational fabric of aviation. Ensuring the security of the data and systems that underpin the digital twin is critical to maintaining trust and confidence among stakeholders. Future research should focus on developing advanced cybersecurity frameworks that can protect digital twins from a wide range of threats, including data breaches, cyberattacks, and insider threats. This could involve research into the use of blockchain technology for secure data storage and transmission, as well as the development of AI-driven threat detection and response systems that can identify and mitigate potential security risks in real-time.

The exploration of emerging technologies such as quantum computing and 6G networks presents another important direction for future research in aviation digital twins. Quantum computing, with its ability to process vast amounts of data and solve complex optimization problems at unprecedented speeds, could revolutionize the way digital twins simulate and predict aircraft behavior under a wide range of scenarios. Research into quantum algorithms specifically tailored for digital twin applications could unlock new levels of accuracy and efficiency in predictive analytics and simulation. Similarly, the advent of 6G networks, with their ultra-low latency and high data transfer rates, could enable real-time updates and interactions with digital twins even during flight, opening new possibilities for in-flight diagnostics and decision-making.

Another promising area of future research is the extension of digital twin technology to broader aviation ecosystems. While current research primarily focuses on individual aircraft, future studies could explore the development of digital twins for entire fleets or even airport operations. This could involve the integration of multiple digital twins into a larger, interconnected system that provides a holistic view of aviation operations. Such an approach could enable more efficient fleet management, improved air traffic control, and enhanced coordination among different stakeholders in the aviation ecosystem.

Sustainability and environmental impact are also becoming increasingly important considerations in aviation. Future research could explore how digital twins can be used to optimize aircraft operations for reduced fuel consumption and lower emissions, contributing to the aviation industry’s sustainability goals. This could involve the development of digital twin models that incorporate environmental factors and simulate the impact of different operational strategies on fuel efficiency and the carbon footprint.

As the aviation industry continues to evolve, these research efforts will be crucial in ensuring that digital twin technology can meet the growing demands of safety, efficiency, and sustainability.

5. Conclusions

This paper has presented a comprehensive framework for the implementation of digital twins in aircraft lifecycle management, focusing on the integration of data-driven models to enhance decision-making processes and operational efficiency. The proposed framework represents a significant advancement in the field of aviation technology, offering a holistic approach to managing the complex lifecycle of modern aircraft.

The framework’s strength lies in its integration of cutting-edge technologies such as IoT sensors, big data analytics, machine learning, 6G communication, and cloud computing. By leveraging these technologies, the digital twin ecosystem created can provide real-time, accurate representations of aircraft throughout their operational life. This capability addresses longstanding challenges in the aviation industry, particularly in maintaining accurate digital representations of aircraft during their operational and maintenance phases.

A key contribution of this work is the detailed exploration of the framework’s components including lifecycle phases, new technologies, and models for digital twins. This paper has demonstrated how these components interact to create a dynamic, evolving digital representation that can adapt to the changing conditions and requirements of an aircraft throughout its lifecycle.

The incorporation of physics-based, data-driven, and hybrid models provides a robust foundation for simulating and predicting aircraft behavior under various conditions. This multi-model approach ensures that the digital twin can accurately represent both the known physical principles governing aircraft operation and the complex, data-driven patterns that emerge from real-world usage.

The supporting components of the framework, including data management and integration, federated learning, and analytics tools, ensure that the digital twin remains a practical, implementable solution. These components address critical issues such as data privacy, collaborative learning, and the generation of actionable insights from vast amounts of data.

This paper has also highlighted the importance of decision-making models within the digital twin framework. By integrating predictive maintenance, operational optimization, and other decision support tools, the framework ensures that the insights generated by the digital twin can be effectively translated into concrete actions that improve aircraft safety, efficiency, and performance.

The exploration of a knowledge-driven approach to digital twins represents a forward-looking aspect of this work. By incorporating diverse knowledge sources and leveraging semantic technologies, this approach paves the way for more intelligent, context-aware digital twins that can provide deeper insights and more nuanced decision support.

While acknowledging the significant potential of this framework, this paper has also addressed the challenges and limitations in implementing aircraft digital twins. Issues such as data quality, integration complexity, scalability, and security remain important areas for ongoing research and development.

Looking to the future, this work has identified several promising directions for further research, including advancements in data integration techniques, more sophisticated predictive analytics, improved AI interpretability, enhanced cybersecurity measures, and the exploration of emerging technologies such as quantum computing.

Digital twins represent a powerful tool in the future of aircraft lifecycle management. By overcoming the current challenges and embracing ongoing technological advancements, the full potential of digital twins can be realized in the direction of more efficient and reliable aviation industry.