Research on a Real-Time Control System for Discrete Factories Based on Digital Twin Technology

Abstract

Gear factories are most typical discrete manufacturing factories. Many gear factories are striving to explore how to achieve intelligent manufacturing in order to improve efficiency and reduce costs. Digital twin technology is currently one of the most reliable ways to achieve intelligent manufacturing. This article aims to establish a real-time control system in order to promote intelligent manufacturing for discrete manufacturing factories. Firstly, a model for a digital twin gear factory is put forward based on the characteristics of gear factories, and the composition of a real-time control system for gear factories is clarified. Then, a human–computer interaction architecture for the real-time control system is proposed. The real-time control system consists of three parts as follows: a monitoring module, a virtualizing module, and a controlling module. At work, it appears as a kind of human–machine interaction form with the three following interfaces: a monitoring window, a virtualizing window, and a controlling window. Finally, a gear factory, which is specialized in manufacturing the intermediate shaft dual gear of a new energy vehicle gearbox, develops a set of software for the real-time control system. The prototype software is obtained through some development activities such as 3D MAX and WebGL virtualization modeling and OPC UA and REST communication design.

1. Introduction

The real economy, dominated by the manufacturing industry, determines a country’s economic strength and economic health level [1]. China, despite having developed into a machinery manufacturer of a certain quantity, still has some way to go when compared with some machinery manufacturers of quality and efficiency in the world [2]. Therefore, the Industry 4.0 concept and the Made in China 2025 strategy are highly respected in the transformation and upgrading of the manufacturing industry. Through a Baidu online search, there are two different production methods in the modern manufacturing industry: discrete manufacturing and process manufacturing. Discrete manufacturing is suitable for producing products which are composed of multiple independent components, such as automobiles, electronic products, mechanical equipment, etc. In discrete manufacturing, products are decomposed into multiple parts or components, produced at different processing locations separately, and ultimately assembled into a complete product. Process manufacturing is process-oriented and is suitable for industries with continuous production processes and relatively fixed product forms, such as chemical, refining, steel, and food and beverage enterprises. In process manufacturing, raw materials are transformed into final products through a series of continuous production steps and chemical or physical changes. Most discrete manufacturing enterprises apply multi-product and small-batch production methods, meaning that the control system of discrete enterprises is usually much more complex than that of process-oriented manufacturing enterprises. Mechanical gears are widely used as mechanical components, and gear manufacturing enterprises are most typical discrete manufacturing enterprises and can serve as the representatives of discrete manufacturing enterprises. Gear factories are the main departments for each gear manufacturing enterprise, and many gear factories are presently striving to promote intelligent manufacturing to improve efficiency and reduce costs. Digital twin technology is currently one of the most reliable ways for gear factories to achieve intelligent manufacturing. Meanwhile, a gear factory is willing to try to use digital twin technology to establish a real-time control system. The gear factory is specialized in manufacturing the intermediate shaft dual gear of a new energy vehicle gearbox, and has currently implemented the digital management of barcodes and QR codes. Now, it wants to transform and upgrade to intelligent manufacturing, but it does not know where to start. Therefore, the research on the real-time control system of discrete factories not only conforms to the Industry 4.0 concept and the Made in China 2025 strategy, but can also meet the needs of transforming and upgrading discrete factories.

As the production mode of the manufacturing industry gradually develops towards complex, integrated, and composite directions, various process data have increased dramatically. Meanwhile, the digital equipment has also been introduced to the manufacturing site, with huge amounts of data reflecting the status of production. Product manufacturing and process control will gradually enter the stage of digital management. In the research regarding digital factories at the present stage, Talkhestani et al. discussed the various existing architecture definitions of digital twins, and elaborated on the three major characteristics of intelligent digital twins as follows: synchronization with the real world, active data acquisition ability, and integration with artificial intelligence, proposing a collaborative simulation development method for agent-based digital twin systems [3]. Jones et al. systematically reviewed the thematic analysis of digital twin technology over the past decade and summarized the characteristics and future research areas of digital twin systems. When describing the characteristics of digital twins, a summary analysis was conducted concerning the conceptual state and key terms [4]. Caesar et al. described the information model of digital twins and introduced the advantages and potential of applying digital twin technology to part processing [5]. There are still many problems with flexibility, dynamics, complexity, data analysis, and other aspects in production management. There is a lack of real-time data perception and acquisition ability, no process-related database established, and no conditions for analyzing and making decisions for big data in order to control and optimize products in each stage.

In 2011, Professor Michael proposed the concept of the digital twin [6]. Digital twin technology is a simulation method, consisting of integrating multiple physical quantity, multiple scale, and multiple probability models, in which historical data and real-time updated data are used to depict and reflect the production process [7]. In 2016, Schroeder et al. first applied the digital twin technology to the industrial area [8]. Morse et al. applied the digital twin technology to the simulation of production design and processing for improving the management and control of products [9]. Tao et al. proposed the concept of the digital factory and designed the five-dimensional model framework of the digital factory in order to study and analyze the main composition [10]. Schluse et al. applied the digital twin to research system modeling and process simulation with the aim of reducing the complexity of the simulation process [11,12]. The five-dimensional model of the digital twin, as well as relevant application areas, were also proposed [13]. Ma et al. designed and constructed instrument- and equipment-sharing laboratories using digital twin technology [14]. Chen et al. proposed a three-dimensional detection algorithm based on digital twin technology, establishing the digital mapping between physical space and virtual space, thus solving the problem of pseudo correspondence in the traditional methods [15]. Nguyen et al. designed a unified real-time ethernet communication structure based on ethernet standards, solving the real-time processing problem of workshop networks [16]. Akerman et al. deployed LTE (Long Term Evolution) networks in the workshop using 5G technology, achieving real-time production data [17]. Latif et al. defined exchange data structures based on OPC UA (OLE (Object Linking and Embedding) for Process Control Unified Architecture) standards and used them for factory data exchanges [18]. Mathias et al. achieved cross network multi-database engine interconnection through two different OPC UA communication methods, effectively solving the problem of connecting different data and views [19]. Otrebski et al. proposed a communication framework using an agent-based architecture for distributed manufacturing systems [20]. The authors of this article were committed to applying digital twin technology to quality management [21] and process control [22], as well as research concerning visualization software for digital twin gear factories [23]. This fully reflects that the research regarding a real-time control system for discrete factories based on digital twin technology has a solid foundation. This article aims to establish a real-time control system for the dual gear factory. The current application status of digital twin technology is investigated. A model for the digital twin gear factory is put forward based on the characteristics of gear factories, and the composition of a real-time control system for gear factories is clarified. Then, a human–computer interaction architecture for the real-time control system is proposed in order to establish the system software functional structure. Finally, the dual gear factory is applied to develop a set of software for the real-time control system.

2. Methods

According to the introduction above, the gear factory is a typical type of discrete factory. So, the research on the real-time control system for a gear factory based on digital twin technology also belongs to the scope of the research concerning the real-time control system for the discrete factory based on digital twin technology. A real-time control system consists of the hardware of the real-time control system and the software of the real-time control system. The hardware includes some objects from a real-time control system, such as sensors, controllers, computers, camera lenses, cable connections, etc. The software includes some other components from a real-time control system, such as computer applications, surveillance videos, images and data, databases, etc. According to the concept of digital twins, a digital twin factory consists of a physical factory and a digital factory. The physical factory includes the hardware of the digital twin factory and the software that is not related to the digital factory. The digital factory includes the software, except for some specialized software which is unrelated to the digital factory, from the digital twin factory. So, the digital factory is simply the software from the real-time control system of the digital twin factory. Due to the extremely simple hardware of the real-time control system and its standardization, the research on the real-time control system mainly focuses on the software of a real-time control system or the digital factory. And the sequence of the research on the real-time control system is as follow: the modeling of the digital twin factory, the design of interaction architecture, and the development of the system software.

2.1. Modeling of Digital Twin Factory

Referring to the current research status of twin factories [21] and combining the operation characteristics of gear factories [22], a model for the digital twin gear factory is designed as a three-dimensional framework based on a database, as shown in Figure 1. The model consists of a database and the three following parts: a physical factory, a virtual factory, and a management system. The physical factory is constructed based on the process characteristics of the products. The virtual factory is generated through mapping, fusion, computation, and reasoning. The management system is established through sensor collection and human–computer interaction, and all parts of the digital twin factory are organically interconnected. The model is used to realize the integration of elements and the fusion of processes for the physical factory, virtual factory, and management system through the entity mapping and real-time interaction of the physical factory and the virtual factory. The physical factory has perceptual access capabilities and can be considered as a type of factory, adding multisource heterogeneous real-time data on the basis of the initial factory, which consists of machinery equipment, personnel, materials, and other components. The virtual factory is a virtual mapping of the physical factory using geometry, behavior, rules, and other models. And it is a highly restored virtual model of the physical factory simulation software. The management system is responsible for the control and optimization of the production elements, production plan, production process, and other aspects, with the aim of improving the production efficiency of the factory. The data of the physical factory, virtual factory, and management system, and the data of their fusion constitute the digital twin data, which eliminates the information isolated island. The real-time prediction, feedback, and monitoring of the production process are achieved via realizing the interactive fusion of the physical space and information space of the process.

The model of the digital twin gear factory is improved based on previous research findings [21]. To achieve real-time control and efficient management, computer networks, camera monitoring, and data analysis should also be included in the digital twin factory. The digital twin factory will be more extensive and advanced than the initial factory.

2.2. Design of the Interaction Architecture

The model for the digital twin factory includes a physical factory, a virtual factory, and a management system, according to the aforementioned viewpoint. However, a gear factory mainly includes personnel, machine groups, processes, materials, environments, and related parameter data, as well as entities such as sensor transmission and reception, data integration, and interaction, with a wide range of multiple components and poor visualization. In order to realize real-time control and efficient management, it is necessary to install some surveillance cameras to monitor the physical factory on a computer screen. The virtual factory needs to include element simulation and process prediction, and the management system should include data exchange and real-time control. The digital twin factory is the equivalent of adding a computer network and a set of computer software to the initial factory. The hardware of the computer network belongs to the physical factory, while the software belongs to the digital factory. And the digital twin factory can also be regarded as a combination, assembled via adding a computer network containing the digital factory on the basis of the initial gear factory. The physical factory is the hardware of the initial entity factory and computer network. The digital factory is the software of the initial entity factory and computer network, and it exists in the form of computer software and consists of a monitoring system, a virtualizing system, and a controlling system. And it is simply the software of a real-time control system for the gear factory based on digital twin technology. The physical factory with the virtual factory, not with the digital factory, is a twin factory. At work, the digital factory appears as a human–machine interface with the three following parts: a monitoring window, a virtualizing window, and a controlling window. A real-time control architecture for the digital twin factory is designed, as shown in Figure 2.

The digital factory includes three parts as follows: a monitoring system, a virtualizing system, and a controlling system, with three split windows as follows: a monitoring window, a virtualizing window, and a controlling window. The monitoring window is used to locate and monitor the physical factory. The analog signals or the digital signals can be used to achieve the real-time monitoring of the physical factory. The virtualizing window is used for virtualizing and simulating the physical factory, as well as predicting the results of various processes. The virtual factory is generated by the virtualizing system, and it is an important part of the virtualization system. The controlling window is used for the analysis and forecast. Each part is organically interconnected and interacts in real time.

2.3. Development of System Software

The software development of a real-time control system is much more complex than that of a general management system. The key links include the following: scene monitoring design, element/process virtualization, a data controlling design, and a system communication design.

2.3.1. Scene Monitoring Design

The gear factory has a considerable amount of equipment and a large site, requiring the installation of the monitoring equipment to achieve real-time control and efficient management. It is necessary to install monitoring cameras and other sensors in the physical factory and connect them to the computer in order to monitor the scene of the physical factory on the computer screen. The signal transmitted from the physical factory to the monitoring window can be an analog signal or a digital signal. The scene monitoring window can interact with the virtualizing windows and the controlling windows for data transmission.

Scene monitoring and servo positioning are mainly completed using the monitoring window. In this window, some pictures from the prestored digital library will be displayed, and the prestored recordings and the real-time captured videos will be played. The displayed contents can be a remote bird’s-eye view of the physical gear factory, a layout diagram of the physical gear factory site, or close-up shots from different perspectives. And these can also be some prestored images from digital libraries, or recorded and real-time captured videos. The displayed contents of the monitoring window can be linked to the displayed contents of the virtualizing window and of the controlling window. The monitoring window display can be seen as a digital abbreviation for the physical gear factory and its local areas, and implementing the monitoring window function is relatively simple.

2.3.2. Element/Process Virtualization

The visualization of the physical factory mainly includes the establishment of basic geometric model libraries, scene constructions, action predefinition, scene optimization, and human–machine interactions [23], as shown in Figure 3. The geometric model library includes a collection of equipment, such as machining devices, logistics and warehousing facilities, and testing units. The scene construction is mainly reflected in the environmental elements, using lighting, special effects, and other methods to simulate the actual physical environment of the gear factory. Action predefinition is the characterization of the different states of the equipment in the production process, thus improving the equipment action attributes and making them relate more closely to the production reality. Scene optimization is the process of enhancing the smoothness, balance, and visibility of a system through rendering, lightweight processing, and other methods. Human–computer interaction refers to the ability of users to roam virtual scenes from the first-person perspective, and the system can receive external commands in order to guide the physical factory production, thus achieving virtual control over reality.

The virtualization of elements and processes is mainly completed using virtualizing window. In this window, some images from the prestored digital library will be displayed, and the prestored virtual images and the process prediction simulation videos will be played. In order to generate a virtual gear factory or achieve predictive simulation displayed in a virtualizing window, the static resources and dynamic processes of the physical factory need to be processed through a series of complex algorithms and programs, such as mapping, fusion, calculation, simulation, and update. Sometimes, it is also necessary to use it in conjunction with the monitoring window. The physical factory is a multifaceted and complex scene, mainly divided into the process equipment, auxiliary equipment, and production material. Below are the virtualization methods for products, devices, and so on.

• Virtualization software tools

The 3D MAX (2015–2024, ten versions) is a 3D-modeling software from the Autodesk Corporation, and the WebGL (Web Graphics Library) is a 3D-drawing technology; the combination of the two can effectively achieve the construction of a virtual factory. By using HTML and JavaScript languages without the need for other plugins, interactive 3D visualization can be achieved on the web side. The physical factory mainly includes various types of equipment, product logistics, etc. The 3D MAX software can be used to build models through polygon modeling, 2D to 3D conversion, and surface modeling. And different modeling methods can be flexibly combined during the modeling process, effectively reducing the modeling workload.

2.

Product virtualization methods

Taking a kind of dual gear as the product to be virtualized, the dual gear is the intermediate shaft dual gear found in a new energy vehicle gearbox, as shown in Figure 4.

The dual gear is a type of double gear assembly made up of a shaft gear and a disc gear pressed together. The manufacturing process flow of the shaft gear is shown as: gear blank, precision turning, spline rolling, hole drilling, gear hobbing, gear burring, heat treatment, laser marking, hole grinding, end turning, cylindrical grinding, gear grinding, spline inspecting, crack detecting, gear cleaning. The manufacturing process flow of the disc gear is shown as: gear blank, spline broaching, gear hobbing, gear burring, heat treatment, laser marking, end turning, spline inspecting, gear cleaning. And the manufacturing process flow of the dual gear is press fitting, laser marking, gear grinding, bite inspecting, gear cleaning, product packaging. Applying 3D MAX and WebGL technologies and defining driving variables such as product location, status, process for virtualization, the product state models can be obtained as shown in Figure 5.

In the figure, the models, respectively represent gear blank, extruded spline, deep hole, hobbing gear, gear blank, pulling keyhole, hobbing gear, finish product. These products can achieve translation, rotation, and scaling through subsequent process virtualization.

3.

Device virtualization methods

The main devices for producing dual gears include broaching machines, gear hobbing machines, milling machines, laser-marking machines, CNC lathes, cleaning machines, tooth-rolling machines, deep-hole-drilling machines, hole-grinding machines, cylindrical grinders, gear grinding machines, press-fitting machines, stackers, and AGVs. A kind of cleaning device, which is used to clean the finished product, is shown in the left half of Figure 6. By applying some 3D MAX and WebGL virtualization techniques, such as mapping and fusion, to virtualize the device, a virtual device was obtained, as shown in the right half of Figure 6.

4.

Scene virtualization methods

The annual output of the dual gear factory is 100,000 pairs of dual gears. It is necessary to sort out and improve the initial entity dual gear factory according to the requirements of digital twin technology. The physical factory is reconstructed to adapt to the digital factory after fully understanding the drawings and processes of automotive and motorcycle gears, and deeply analyzing the product features, production quantity, process route, service department, delivery time, as well as factory facility planning, equipment selection, production layout, and management design. The overall layout of the factory is then organized and drawn, as shown in Figure 7.

When virtualizing the dual gear factory based on the 3D MAX and WebGL technologies, the physical factory should be divided into components and blocks. Some simulation techniques such as mapping and fusion are applied in order to virtualize the components and blocks, and simulation models are built for material products, production devices, facility environments, and production processes to achieve the virtualization of the physical factory.

Corresponding to the physical dual gear factory in Figure 8 and following a series of modeling, calculating, programming, and other operations, a virtual dual gear factory is generated, as shown in Figure 8.

5.

Process virtualization methods

The production activities in the virtual factory can usually be composed of methods such as translation, rotation, and scaling. If the spatial position coordinate vector of the virtualization model is assumed to be (and the reference ratio, x y z r) and the reference ratio r = 1, the transformation expression for translation, rotation, and scaling is as follows:

$$\left[x^{\prime } y^{\prime } z^{\prime } 1\right]=\left[x y z 1\right]\left[\begin{array}{cccc}a& b& c& d\\ e& f& g& h\\ i& j& k& l\\ m& n& o& p\end{array}\right]$$

(1)

In the formula, (x y z 1) is the original coordinate vector being transformed, and (x′ y′ z′ 1) is the transformed coordinate vector. By changing this fourth-order transformation matrix, the translation transformation, rotation transformation, and scaling transformation of the virtual model can be achieved.

The translation transformation expression is shown in Equation (2).

$$\left[x^{\prime } y^{\prime } z^{\prime } 1\right]=\left[x y z 1\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& 0\\ p_x& p_y& p_z& 1\end{array}\right]$$

(2)

Through this expression, the virtual model has translated Px, Py, and Pz distances along the X, Y, and Z axes, respectively. And, when the virtual model rotates around the X, Y, and Z axes, respectively θ, the rotated coordinates can be expressed as Equations (3)–(5).

$$\left[x^{\prime } y^{\prime } z^{\prime } 1\right]=\left[x y z 1\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& cos\theta & sin\theta & 0\\ 0& -sin\theta & cos\theta & 0\\ 0& 0& 0& 1\end{array}\right]$$

(3)

$$\left[x^{\prime } y^{\prime } z^{\prime } 1\right]=\left[x y z 1\right]\left[\begin{array}{cccc}cos\theta & 0& -sin\theta & 0\\ 0& 1& 0& 0\\ sin\theta & 0& cos\theta & 0\\ 0& 0& 0& 1\end{array}\right]$$

(4)

$$\left[x^{\prime } y^{\prime } z^{\prime } 1\right]=\left[x y z 1\right]\left[\begin{array}{cccc}cos\theta & sin\theta & 0& 0\\ -sin\theta & cos\theta & 0& 0\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]$$

(5)

When the virtual model is scaled proportionally along the coordinate axis, the transformation expression can be shown using Equation (6).

$$\left[x^{\prime } y^{\prime } z^{\prime } 1\right]=\left[x y z 1\right]\left[\begin{array}{cccc}{s}_x& 0& 0& 0\\ 0& s_y& 0& 0\\ 0& 0& s_z& 0\\ 0& 0& 0& 1\end{array}\right]$$

(6)

In the formula, Sx, Sy, and Sz are the scaling ratios of the virtual model on the X, Y, and Z axes, respectively.

The description of all device actions can be achieved through transformation and combination processes. After importing the model, the model actions can be decomposed into translation, rotation, and scaling. Through defining the action controllers encapsulated by WebGL technology, and by setting the action type and the motion distance, respectively, the action controllers are bound to the model to control the model actions using driving variables.

Additionally, the scene-roaming mode can be used to enhance the user experience. By simulating the perspective of the user walking in the scene, the user can obtain a feeling of on-site observation. In this mode, users can use the directional keys to freely move in the scene and interact with each other by moving to the designated location according to their needs. If the user wants to obtain the processing status of the relevant equipment, they can click on the corresponding virtual model with the mouse, and detailed information about the equipment’s operation will appear next to the model.

2.3.3. System Controlling Design

After achieving the unified integration of factory data communication, it is necessary to establish a factory controlling method. The focus of control is on the data analyzing system. The data analyzing system is used to analyze multiple sources of data in order to obtain factory status information, and to achieve factory anomaly control and continuous improvement. Data analysis can establish a control index system based on axiomatic design and functional requirements. The quality function deployment method can be used to expand order customer needs into production service capability requirements. A quality house can be established to obtain the correlation between the production service capability and the control indicators. The weight of control indicators can be calculated based on the importance of the production service capability, achieving the quantitative evaluation of the factory control capability. Simultaneously, establish a real-time data-driven control process, using a grey Markov prediction model and anomaly pattern recognition models in order to achieve process data prediction and anomaly recognition. Figure 9 shows a set of control charts, histograms, and proportion charts for the grinding quality analysis of the dual gear shaft shoulder presented in the analyzing window.

The graphic and text database stores some virtual modeling data and basic components; modeling algorithms and many applications are developed used artificial intelligence technologies such as axiom inference, grey theory, ant colony algorithm, and machine learning. Data analysis and application development are mainly completed using the analyzing window. In this window human–computer interaction, the interface is mainly displayed. It is used for software development, functional activation, data input, parameter modification, process simulation, result prediction, as well as for adjusting and controlling the operational status of physical gear factories. The analyzing window determines the display status of the monitoring window and the virtualizing window.

2.3.4. Data Communication Design

After building a virtual factory, in order to achieve the real-time driving and system data integration of the virtual factory, communication technology is needed to achieve system data interconnection and interoperability. By analyzing the current status of factory communication and the characteristics of digital twin technology, the OPC UA technology and REST (Representational State Transfer) architecture are used to achieve unified data exchange between physical factories and control systems [23]. The OPC UA technology, which is a unified architecture based on object linking and embedding for process control, can be applied to achieve data transmission between the physical factory and the digital factory, or rather between different devices and operating systems, simultaneously in order to integrate multisource heterogeneous data and enhance the availability of the digital factory. The process of constructing OPC UA communication can largely be divided into two steps. The first step is to treat factory elements as objects, abstract attributes, and behaviors, and to establish an information model. The OPC UA technology uses a unified information model to describe objects, which is represented by an address space, and the nodes are the basic units of the address space. The OPC UA identifies the connection semantics between nodes through different types of references, forming network association relationships within the address space. The second step is to convert the information model description file into programming code, and then use the OPC UA server to adapt the transmission protocol to complete data access for different types of devices. External applications can access the data interface provided by the OPC UA server, and data reading and writing can be achieved by manipulating variable node attribute values through node addresses. Drawing on object-oriented thinking, the physical factory is divided into categories and is encapsulated into nodes, with each node referencing each other. For example, mapping properties, such as data and activities, to objects, variables, and methods, by referencing types through different components and properties, and by using XML files to describe the information model of static resources, the information communication can be achieved based on the OPC UA technology.

REST is the abbreviation for representational state transfer. The REST architecture is a new style of internet software architecture, which enables software or applications to transmit information in different environments. It has excellent compatibility and is suitable for applications in multi-system complex service invocation environments. The methods of interface definition and data representation are unified based on REST architecture specifications to implement system management. The REST architecture drives interfaces using the HTTP protocol, providing a unified semantic operating standard for all resource interfaces. The HTTP protocol is not only a data transfer protocol, but also a state transition protocol. The REST architecture uses HTTP verbs to distinguish resource operation types, and uses GET, POST, PUT, and Delete to represent resource acquisition, addition, modification, and deletion operations, respectively. This allows a data interface to have four operation types, thus greatly reducing the number of interfaces and making interface calls more concise. The REST architecture uses a JSON data format to unify data representation and avoid data parsing errors [23]. When compared with other forms of data representation, it displays the characteristics of simple data formats, easy parsing, and cross programming languages, especially the cross programming language characteristics, which are very suitable for use in complex system environments. Through the hard work of multiple graduate students, data transmission and interaction between the physical factory and the digital factory, and among three split windows, have been successfully achieved using the two standard frameworks.

3. Results

Through the innovative research and extensive programming mentioned above, a prototype software of the real-time control system, also known as a digital factory, has basically taken shape. It is made into a three-split window human–machine interaction system for the digital twin dual gear factory, as shown in Figure 10. And the functions of each split window are clarified [23].

In the digital twin dual gear factory, the physical factory is the initial entity factory and the hardware of the computer network. And the digital factory is the computer software. The virtual factory is a part of the digital factory. The digital factory consists of the monitoring window, the virtualizing window, and the controlling window. Three windows can be located on the same computer screen, and each window can also be individually enlarged to cover the full screen. The physical scene, virtual factory, and analyzing program of the digital twin factory are displayed in three windows, using the aforementioned virtualization software, communication technology, and data analysis methods. Figure 4, Figure 5, Figure 6, Figure 8 and Figure 9, used to describe software development techniques, were also generated using the three-window system. Although only a small number of functions for monitoring, virtualizing, and controlling can be achieved, and also need to be further improved, the failure to make the software more powerful and mature comes as a result of the authors’ insufficient understanding of the prototype factory and excessive software development workload, and is not due to technical reasons. Finally, the feasibility and applicability of the real-time control system for the digital twin dual gear factory proposed is indirectly verified by some images being displayed in the three windows.

4. Discussion

The final research result of this article is a prototype software of the real-time control system for the factory, which is specialized in manufacturing the intermediate shaft dual gear of a new energy vehicle gearbox. But the final result cannot be separated from the previous processes and methods, and discussing the final outcome will also involve the previous processes and methods.

• The relationship between the final research result and title of this article: The intermediate shaft dual gear of a new energy vehicle gearbox is a type of mechanical gear, and mechanical gears are widely used as mechanical components. Gear manufacturing enterprises are most typical discrete manufacturing enterprises. The factory is the main department for each manufacturing enterprise. A gear factory is an abbreviation for a gear manufacturing factory, and a discrete factory is an abbreviation for a discrete manufacturing factory. The gear factory is a most typical discrete factory, and the research methods and achievements of a real-time control system based on digital twin technology are also suitable for other discrete factories.
• The relationship between the digital factory and a real-time control system: A digital twin factory consists of a physical factory and a digital factory. The physical factory includes the hardware of a digital twin factory and the software that is not related to the digital factory. The digital factory includes the software, except for some specialized software which is unrelated to the digital factory, of the digital twin factory. Then, a real-time control system consists of the hardware of the real-time control system and the software of the real-time control system. The hardware includes some objects of a real-time control system, such as sensors, controllers, computers, camera lenses, cable connections, etc. The software includes other components of a real-time control system, such as computer applications, surveillance videos, images and data, databases, etc. Therefore, the digital factory is just the software of the real-time control system of the digital twin factory. Due to the extremely simple hardware of the real-time control system and its standardization, the research on the real-time control system mainly focuses on the software of a real-time control system or the digital factory. The software development of a real-time control system can be transformed into the development of a digital factory, laying an important technical path for researchers studying real-time control systems.
• The impact of the model for a digital twin gear factory on the prototype software of the real-time control system for the dual gear factory: the model for a digital twin gear factory is the centralized display of software and hardware resources, as well as the operating methods for the digital twin gear factory. And it determines the functionality and applicability of the prototype software of the real-time control system, and also affects the requirement analysis and functional implementation of software development.
• The impact of the human–computer interaction architecture for the real-time control system on the prototype software of the real-time control system for the dual gear factory: The human–computer interaction architecture mainly affects the performance and operation of software, and also has a certain impact on the requirement analysis and functional implementation of software development.
• The impact of the virtual tools, such as 3D MAX and WebGL, and communication standards, such as OPC UA and REST, on the prototype software of the real-time control system for the dual gear factory: The selection and use of virtual tools affect the quality of THE virtualized images and development efficiency, while the selection and use of communication standards affect system data transmission and interactions. The reasonable combination of 3D MAX and WebGL can basically meet the requirements of virtualization development, and the reasonable combination of OPC UA and REST can also basically achieve the requirements of data communication.
• Reasons for imperfect and immature software system development: A prototype software of the real-time control system for the factory, which is specialized in manufacturing the intermediate shaft dual gear for a new energy vehicle gearbox, is developed. But only a small number of functions for monitoring, virtualizing, and controlling can be achieved and need to be further improved. The failure to make the software of the real-time control system more powerful and mature is not due to technical reasons, but rather the authors’ insufficient understanding of the prototype factory and excessive software development workload.
• However, only the physical geometric properties and basic processing actions of the virtual factory are researched in this article, and the factory operation logic and processing trajectory have not yet been considered. More realistic methods need to be further studied and improved. The practice of digital twin factories is still in the exploratory stage, and there are still many key issues that need to be further studied and improved in the implementation process. In virtual factory modeling, the virtual factory needs to reflect the real production scene, which places high demands on modeling techniques. The integration control of the factory prototype system only provides critical control for the digital twin factory, and the provided control is not comprehensive enough to integrate other management systems. In the future, more artificial intelligence and big data technologies will be combined to achieve more levels of data analysis work.

5. Conclusions

In this article, the relationship between the gear factory and the discrete factory is clarified. The relevant research results and the current application status of digital twin technology are analyzed. The real-time control system of a gear factory and the composition of a digital twin gear factory are compared. The software architecture of a real-time control system is researched. A digital twin gear factory is established, and a prototype software of the real-time control system is obtained. The final conclusions are as follows:

• A gear factory is a typical discrete manufacturing factory, and the research methods and achievements of a real-time control system for the dual gear factory based on digital twin technology are also suitable for other discrete factories.
• The digital factory is just the software of the real-time control system of the digital twin factory. The research on a real-time control system mainly focuses on the software of a real-time control system or focuses on the digital factory.
• A model for the digital twin gear factory is put forward based on the characteristics of gear factories. This model can transform the software development of a real-time control system into the development of a digital factory, laying an important technical path for researchers studying real-time control systems.
• A human–computer interaction architecture for the real-time control system is put forward. It consists of three parts as follows: a monitoring module, a virtualizing module, and a controlling module. At work, it appears as three interactive interfaces as follows: a monitoring window, a virtualizing window, and a controlling window. A template for the software development of a real-time control system is proposed.
• The virtual factory and some visualizing processes are created using 3D MAX and WebGL technology, and the device communication and interface management are implemented with OPC UA and REST standards. These technologies and standards can improve the software development efficiency of a real-time control system, thereby accelerating the development of virtual technology.
• A prototype software of the real-time control system for the factory, which is specialized in manufacturing the intermediate shaft dual gear of a new energy vehicle gearbox, is developed. But only a small number of functions for monitoring, virtualizing, and controlling can be achieved and need to be further improved. The failure to make the software of the real-time control system more powerful and mature is not due to technical reasons, but rather the authors’ insufficient understanding of the prototype factory and excessive software development workload.