Leveraging Variable Frequency Drive Data for Nondestructive Testing and Predictive Maintenance in Industrial Systems

Abstract

Nondestructive testing (NDT) has a crucial role in ensuring the reliability and safety of industrial systems. However, traditional methods typically rely on external sensors, which can lead to increased costs and added complexity. The current study examined an alternative approach using variable-frequency drive (VFD) data for real-time fault detection and predictive maintenance. Most VFDs continuously monitor essential parameters such as motor speed, torque, efficiency, and power consumption, facilitating sensorless condition monitoring that helps detect early-stage motor and apparatus faults without additional hardware. To improve diagnostic capabilities, calculated metrics such as apparent power, efficiency, torque, and energy consumption can deliver more profound insights into system performance, assisting in identifying potential failure patterns. A Python-based data acquisition and visualization system was developed and implemented as an example of a potential solution, enabling centralized monitoring, anomaly detection, and historical data analysis. Future advancements in artificial intelligence and machine learning could further refine automated fault detection by utilizing historical VFD data to predict system failures accurately. By integrating VFD-based diagnostics into NDT, industries can develop scalable, cost-effective, intelligent testing and maintenance solutions that improve reliability and asset management in modern systems.

1. Introduction

Nondestructive testing (NDT) includes various techniques to evaluate the properties of materials, components, and systems without causing damage [1,2]. For practical purposes, the advantages of NDT involve defect detection, predictive maintenance, quality assurance, safety enhancements, and cost discipline [1]. Traditional NDT methods rely on established techniques such as dye penetrant testing, magnetic particle inspection, eddy current testing, radiography, ultrasonics, and acoustic emission analysis [2]. Additionally, many NDT applications utilize external sensors for precision, defect detection, and assessing environmental changes [3].

Nondestructive testing is extensively utilized across various industries where safety during development and mechanical integrity are critical. In aerospace, it identifies material defects and fatigue in aircraft components, ensuring specified reliability [2]. In the oil and gas sector, NDT helps prevent pipeline failures by monitoring corrosion and material degradation [4]. The manufacturing industry relies on NDT for weld inspections and defect detection in castings, enhancing product quality and compliance [1]. In power generation, including nuclear and renewable energy, NDT is employed to assess turbines and high-pressure systems [5]. However, these conventional NDT techniques often lead to higher costs and increased system complexity. While effective, these methods require additional infrastructure and regular inspections, frequently hindering continuous data capture.

With the growth of industrial automation and predictive maintenance strategies, diagnostic data from variable frequency drives (VFDs) have become a valuable yet underutilized resource in NDT. Variable frequency drives are designed to monitor critical motor and apparatus parameters internally and online for self-protection and enhanced performance, thus serving as an inbuilt repository for external data [6]. When VFD data are exported and utilized even for simple condition monitoring, they can provide real-time, continuous surveillance of system health with minimal reliance on external sensors, reduced costs, and decreased complexity [7].

This technical note addresses the underutilized role of variable frequency drives (VFDs) in enhancing nondestructive testing (NDT) and diagnostics to improve the reliability and safety of machinery, materials, and components. Rather than relying on only traditional NDT methods that often require periodic checks and manual application, VFDs provide real-time operational data to help detect anomalies during operation [2,8]. Furthermore, VFD data can be integrated into machine learning (ML) and artificial intelligence (AI) models to predict failures and optimize maintenance strategies using archived data [6,9]. Importantly, VFD-based diagnostics do not aim to replace conventional NDT methods but rather serve as a complementary tool to enhance current data acquisition and support NDT efforts [10]. By combining the information obtained from VFDs with traditional NDT techniques, users can shift toward more predictive, data-driven approaches and maximize the advantages of NDT applied across industries.

2. VFD Data for NDT Applications

The primary function of a VFD is to control the speed of electric induction motors (or variations of permanent magnetic motors) across their entire operating range [11]. In its simplest form, a VFD manages three key output characteristics: the fundamental frequency and voltage magnitude, which maintain a fixed ratio to produce torque in the motor, and the carrier frequency, which determines the quality of switching operations, typically at a minimum of 2 kHz [11]. Digital processing capabilities have been integrated into modern VFD designs since their inception, allowing precise motor control as well as the collection and storage of diagnostic data [7,11]. These data facilitate the monitoring of the drive itself, the power supply, the electric motor, and even the attached load [7]. While very few parameters of early VFDs were available for export for reliability analysis and NDT, modern designs offer most of these parameters for the user’s benefit [7,11,12].

One of the most compelling motivations for accessing VFD data lies in their potential for sensorless condition monitoring, particularly for critical industrial equipment such as centrifugal pumps [5]. Pumps are essential in various industries, including pulp and paper production, wastewater treatment, and chemical processing, where reliability and energy efficiency are paramount [5]. As noted, traditional condition monitoring systems rely on external sensors, which increase system complexity and cost [12]. Recent research has demonstrated that motor torque estimation data can be utilized for fault detection, eliminating the need for external sensors and underscoring the flexible and valuable data generated by VFDs [5]. However, a significant challenge in condition monitoring is the scarcity of data indicating a fault or a defective condition requiring action. The literature often highlights deficiencies similar to those of the critical pumps detailed by Turunen et al. [5]. The answer to this data imbalance may potentially be found in the AI and ML techniques discussed later.

Additionally, VFDs have the capability to also expand their data capture through external sensors via analog and discrete programmable inputs akin to traditional techniques but integrated [13]. For example, the ABB ACS880 drive includes three expansion slots, one of which is typically allocated for communication, while the remaining two can support analog option cards, enabling up to four additional 4–20 mA input signals [14]. In practical terms, these inputs can be utilized for vibration, temperature, flow, or other sensor types required by the application, eliminating the need for additional remote input capture devices and simplifying system integration.

For the implementation presented below, using Modbus TCP/IP (noting that the drive manufacturer dictates the version) via Python to communicate with an ABB ACS880 VFD, it was essential first to select critical parameters, define their mapping structure, and establish their relevance to NDT [15,16,17]. Variable frequency drives continuously measure and log key performance indicators that correlate directly with potential failure modes [8,18].

Table 1. Critical VFD diagnostic parameters.

Parameter	Modbus Register Address	Unit
Motor speed	400005	RPM
Motor voltage	400007	VAC
Motor current	400008	Amps
Motor power	400009	kW
Motor frequency	400010	Hz
Motor temperature	400011	C
DC link voltage	400012	VDC
Run time counter	400013	Hours

Note. This table presents the parameters for an ABB (2020) ACS880 and the corresponding Modbus TCP/IP addresses, emphasizing function codes three and 16-bit words. If the drive is not an ABB model or utilizes a different industrial protocol (as discussed in more detail later), the addressing scheme for the parameters will vary.

provides an overview of commonly available VFD parameters that offer insights into motor health and operating conditions.

Table 2. VFD failure mode comparisons.

VFD Parameter	Related Failure Mode	NDT Relevance
Motor current	Rotor eccentricity, load imbalance, overload, and underload	Identifies excess electrical load, mechanical load, and application design quality
Motor power	Overheating and excessive torque	Detects energy losses
DC link (bus) voltage	Power quality issues, phasing, overvoltage, and undervoltage	Identifies voltage transients, load motoring, and load regeneration
Motor speed	Bearing wear, misalignment, load changes, stall, and overhaul	Detects irregular speed patterns

Note. The table introduces related failure modes and the relevance of NDT; however, the list and interpretation are subject to the application.

illustrates how these parameters align with NDT methodologies.

3. Data Transformation and Learning

The integration of calculated metrics such as apparent power, efficiency, torque, and energy consumption expands the versatility of VFD data beyond basic motor control [11]. While raw parameters such as voltage, current, and speed provide valuable operational insights, calculated values offer deeper system-level diagnostics [6]. For instance, apparent power (kVA) and efficiency (%) can be used to assess energy performance, and torque estimation can be used for mechanical load analysis as shown in

Table 3. Examples of calculated metrics.

Calculated Metric	Formula
Apparent power (kVA)	$Apparent power={\displaystyle \frac{V\times I\times ∛}{1000}}$
Efficiency (%)	$Efficiency=\left( {\displaystyle \frac{Power output\left(kW\right)}{Apparent power\left(kVA\right)}}\right)\times 100$
Torque (Nm)	$Torque={\displaystyle \frac{Power output\left(kW\right)\times 9500}{Motor speed \left(RPM\right)}}$
Energy consumption (kWh)	$Consumed=Power output\left(kW\right)\times Run time \left(hours\right)$

and applied in the later Python example [5,6]. Thus, VFD data are not only a snapshot of motor operation but a comprehensive diagnostic tool for fault detection, performance optimization, and predictive maintenance [4].

Rich data, whether in their raw or derived state, present a significant challenge for condition monitoring due to the imbalance between normal operational data, which comprise the majority of collected data and the rare occurrences of abnormal conditions [8,9]. To address this issue, ML, a subset of AI, leverages training data to compare with collected data, enabling pattern recognition through selected algorithms. Models based on anomaly detection algorithms, recurrent neural networks (RNNs), and convolutional neural networks (CNNs) analyze historical VFD datasets to identify failure patterns [8,9]. This data imbalance can be mitigated through techniques such as synthetic data generation or by transferring relevant data from adjacent applications [8,9].

Beyond data transformation, traditional ML classifiers, which include supervised, unsupervised, and semi-supervised models, have a crucial role in predictive maintenance. Decision trees, random forests, and support vector machines (SVMs) are widely used for anomaly detection in industrial systems [6]. Clustering algorithms such as k-means and Gaussian mixture models (GMMs) provide unsupervised methods for identifying abnormal motor behavior, while semi-supervised learning techniques utilize healthy operational data to detect faults in real time, offering solutions when labeled fault data are scarce [8,9].

Deep learning further enhances predictive maintenance and NDT applications by autonomously extracting patterns from high-dimensional sensor data. Unlike traditional ML, deep learning models continuously train on real-time motor and VFD parameters, improving their ability to detect anomalies before they escalate into critical failures [8,9]. While CNNs and RNNs analyze time-series VFD data to detect failure patterns, autoencoders and generative adversarial networks (GANs) enhance anomaly detection in highly imbalanced datasets. A deep learning-based fault diagnosis system for centrifugal pumps, examined by Turunen et al. [5], successfully detected cavitation and metal-to-metal contact using only VFD torque estimation data, demonstrating the untapped potential of drive-based diagnostics. By integrating multimodal sensor data and deploying lightweight deep learning models on edge devices, industries can enhance predictive maintenance strategies, reduce false positives, and optimize equipment reliability.

Beyond fault detection, these expanded metrics reconcile the gap between electrical and mechanical failure analysis in NDT [9]. Rather than relying solely on external sensors, trends in torque fluctuations, power loss, and efficiency drops serve as early indicators of potential failures such as bearing degradation, misalignment, or insulation wear [4,18]. Sensorless fault detection using VFD data has effectively diagnosed cavitation in pumps and metal-to-metal wear in bearings, illustrating how critical failure modes can be identified through drive-derived analytics alone [5]. This approach reduces reliance on additional hardware, making NDT applications more cost-effective and scalable. Furthermore, by archiving and analyzing VFD-calculated metrics over time, industries can leverage ML- and AI-driven diagnostics for pattern recognition and predictive fault detection, further optimizing maintenance strategies [8,9].

4. Accessing VFD Data via Industrial Communication Protocols

The user has many options for how data flow from the inbuilt repository of the VFD. Many systems use programmable logic controllers (PLCs), distributed control systems (DCSs), or proprietary controllers, where the drive can be a simple field device exchanging data or a smart device transforming data [17]. The decision to even capture the data versus just having a threshold warning or fault must also be made. The selection of design can depend on the number of VFDs in the applied system, where a standalone VFD without complementary functions can use a program akin to the supplied Python example [19,20] or an existing system in a paper mill, where hundreds of VFDs are controlled by a PLC or DCS, requiring an advanced strategy of integration or replacement [15,21].

It is important to note that PLCs and DCSs can be the center of both collection and transformation through the ML techniques noted earlier or can serve as a bridge to a tertiary system [16]. Additionally, the expanse of the larger system can involve an enterprise solution as part of an Industry 4.0 design where the VFDs and corresponding control become Industrial Internet of Things (IIoT) field and edge devices [12,22]. The complication has become the interchangeable language used where condition monitoring, AI, ML, Industry 4.0, and IIoT have become inseparably entangled [4,8].

In the noted options for flow and control, a protocol is a standardized communication method that enables devices such as VFDs, PLCs, and DCSs to exchange data efficiently [17]. Unlike traditional hardwired input/output connections, protocols such as Modbus, PROFINET, and EtherNet/IP use digital networks to transmit multiple parameters over a single connection, reducing wiring complexity and improving data accessibility [17]. This shift from point-to-point wiring to fieldbus communication enhances real-time control, diagnostics, and system scalability across industrial applications [21].

Using fieldbus to access VFD data offers several benefits [22]. It enables faster response times because multiple parameters such as motor speed, torque, and fault diagnostics can be transmitted in real-time continuous monitoring and predictive maintenance [15]. Fieldbus also reduces wiring costs by consolidating signals into a single network, minimizing installation complexity [15,17]. Additionally, centralized supervisory control and data acquisition (SCADA) or cloud-based systems can collect, log, and analyze VFD data remotely, improving efficiency in industries such as manufacturing, energy, and wastewater treatment [15,16]. By integrating advanced fieldbus protocols, VFDs can synchronize with other industrial devices, optimizing coordinated control in multidrive systems [12].

Designers select different protocols based on factors such as industry standards, speed requirements, and system architecture [15]. For example, PROFINET is commonly used in Siemens-based automation, while EtherNet/IP is preferred for Rockwell/Allen-Bradley PLCs [21]. High-speed protocols such as EtherCAT and SERCOS III are ideal for precision motion control, whereas Modbus TCP/IP remains a cost-effective, vendor-neutral solution for SCADA and remote monitoring [21]. Scalability must also be considered—PROFIBUS and EtherNet/IP support large, complex networks, while CANopen and DeviceNet are used for simpler real-time control [21]. By choosing the correct protocol, industries can ensure seamless integration, optimized performance, and improved reliability in their automation systems (

Table 4. Common industrial communication protocols and use cases.

Protocol	VFD Compatibility	Use Case
Modbus TCP/IP	Standard for most VFDs	Data logging and SCADA systems
PROFINET	Siemens automation systems	High-speed motor control
EtherNet/IP	Rockwell/Allen-Bradley PLCs	Real-time process control
EtherCat	Beckhoff and industrial automation	Ultra-fast cycle times
POWERLINK	B&R and open-source industrial protocol	Deterministic Ethernet for motion control
CANopen	Widely used in automation and robotics	Real-time control in decentralized systems
DeviceNet	Rockwell/Allen-Bradley PLCs	Industrial automation and machine control
SERCOS III	Servo drives and CNC machines (Rockwell)	High-speed motion control
HART	Process automation (ABB, Emerson, and Siemens)	Communication with field devices in hazardous environments
MQTT	Cloud-based Industrial Internet of things	Reliable low-bandwidth data transmission for remote monitoring

Note. The table displays a non-exhaustive list of common protocols to illustrate a cross-section across applications.

; [15]).

5. Implementation: Real-Time Monitoring with Python

The motivation for using Python to communicate with a VFD, as an example, lies in its flexibility, ease of use, and extensive library support for industrial automation [19]. Unlike proprietary PLC or DCS programming languages, Python offers cross-platform compatibility and integrates seamlessly with data analytics, ML, and IIoT cloud-based systems [19]. By leveraging Python or other similar programs, engineers can create custom monitoring solutions that go beyond traditional control logic, enabling the real-time diagnostics, predictive maintenance, and remote monitoring of VFD performance [4]. The prospective design layout is shown in Figure 1.

Using Pymodbus, Python can efficiently read and write VFD parameters over Modbus TCP/IP, eliminating the need for additional hardware interfaces. The ability to log speed, power, current, and fault codes into a CSV file or database provides a structured approach for historical analysis and trend detection. Additionally, Python allows the integration of calculated metrics such as efficiency, torque, and energy consumption, enhancing the depth of diagnostics without requiring external sensors. However, as shown in the following example and the initial Python code snippet (Figure 2), external sensors for additional data (e.g., vibration) can easily be added.

Matplotlib and data visualization capabilities developed by Python provide the benefit of real-time graphical representations of VFD performance [20]. Instead of manually reviewing log files, engineers can utilize live plots to track anomalies, optimize motor performance, and trigger alerts when the parameters deviate from the expected values shown in Figure 2. By integrating data acquisition, computation, and visualization, Python offers a cost-effective, scalable, and highly customizable solution for VFD monitoring and NDT applications [19,20]. Figure 3 shows the Python code that creates the dashboard, as displayed in Figure 4 and Figure 5. The complete Python code is available in a text file at the link provided under Supplementary Materials.

The NDT dashboard and Python code presented serve as the foundation for condition monitoring, enabling the addition of thresholds and alarms based on the monitored application for trending or acute occurrences, together with reliability considerations (as predictive maintenance) such as the remaining useful life (RUL). Integrating RUL estimation into the NDT dashboard enhances predictive maintenance by providing real-time insights into the health of VFD and rotating equipment. By leveraging historical VFD data and degradation trends, the dashboard can calculate RUL using ML models such as linear degradation, regression-based estimation, or deep learning for sequential failure prediction [8,9]. The Python implementation would utilize VFD time-series data—current, failure thresholds, and degradation rates in this case—to compute RUL dynamically [6]. Future research can explore hybrid AI models, combining Weibull analysis, Bayesian networks, and deep learning to improve prediction accuracy and reliability in industrial applications [5]. Additionally, integrating edge computing for RUL estimation would enable real-time analytics directly within IIoT-enabled VFD systems, reducing reliance on cloud processing and enhancing responsiveness in predictive maintenance strategies [4]. An example of this is shown in the code snippet in Figure 6 and Figure 7.

6. Conclusions

Integrating VFD diagnostic data into NDT methodologies enhances fault detection and predictive maintenance without relying on additional external sensors. By leveraging the built-in monitoring capabilities of VFDs, real-time data on motor speed, torque, power, and efficiency can be used to identify early-stage failures such as bearing wear, misalignment, and insulation degradation. This approach not only reduces hardware costs but also streamlines condition monitoring, making predictive diagnostics more accessible and scalable across various industrial sectors.

The real-time monitoring system described in this review and application illustrates how Python-based data acquisition aids in extracting, analyzing, and visualizing vital VFD parameters. With fieldbus protocols such as Modbus TCP/IP, essential drive data can be continuously logged and integrated into SCADA or cloud-based platforms for remote monitoring. Moreover, calculated metrics such as efficiency, torque, and energy consumption offer deeper insights into motor performance and operational health. The key benefits of VFD-integrated NDT include reduced costs due to minimized reliance on external sensors, real-time condition monitoring for continuous diagnostics, scalability across diverse industrial applications, and AI-driven predictive analytics for automated fault detection and maintenance planning.

In addition, AI and ML are promising technologies for improving VFD-based diagnostics. By training models on historical VFD data, AI-driven systems can identify failure patterns before they develop into critical issues, facilitating proactive maintenance strategies. Future research should focus on advancing deep learning techniques for VFD-based anomaly detection, exploring multi-sensor fusion by integrating VFD data with vibration and thermal imaging, and expanding IIoT predictive maintenance frameworks for large-scale industrial applications.

Ultimately, the expanded role of VFD data in NDT signifies a transformative shift in industrial reliability management. By utilizing real-time diagnostics, predictive analytics, and scalable automation, industries can achieve greater efficiency, reduced downtime, and improved equipment longevity. As industrial automation continues to evolve, VFD-integrated NDT solutions will be vital in advancing condition monitoring technologies and ensuring the reliability of essential assets.