Fault diagnosis of electrical equipment based on virtual simulation technology

1 Introduction

There are two methods for the inspection and maintenance of electrical equipment in the production plant: the first is the maintenance method after an accident occurs. This method is to be repaired after the electrical equipment in the installation has a fault problem; in this way of maintenance, the accident of the production device has occurred, and it has an impact on the continuous, stable, and safe production operation of the device; therefore, the economic loss and time loss that the production enterprise has to bear under this maintenance method are also the largest. Since the economic loss of this maintenance method is the largest, in this enterprise, among the requirements for electrical equipment management, the inspection and maintenance work after the failure is only used during the operation of the production plant; the normal production and operation of the production device does not have a substantial impact; the degree is very small; it does not require electrical equipment that must be running at all times, and has spare electrical equipment, or uses other forms of inspection and maintenance methods that are uneconomical and unscientific device [1]. The second maintenance method is preventive regular maintenance. In this method, based on the operating time of the equipment, the maintenance staff usually divide the maintenance of electrical equipment into medium repairs and minor repairs; whether the electrical equipment needs to be overhauled is judged by the sound of the electrical equipment running and the temperature of the equipment during the staff patrol inspection process; this requires employees to summarize their experience, so there will be errors in judgment; especially for new employees with less experience, the probability of errors will increase, for electrical equipment that is not in operation for a long time; because there are established management regulations, employees must follow the content of the management regulations, i.e., no matter what the state of the electrical equipment is in operation, when the time is up, the staff will arrange a plan for maintenance [2]. This preventive maintenance work according to the time factor also has advantages and disadvantages, but with the development of science and technology over the years, the development of the system to high-voltage, large-scale, network technology information, and the improvement of the requirements of the power supply department and the power consumption department, this traditional maintenance method is not suitable for today’s rapidly developing large enterprises; the main reasons are as follows: (i) it is necessary to overhaul and maintain the equipment of the device when the power is cut off, in order to enable the continuous operation of the production device in a safe, stable, and reliable manner, and the important electrical equipment of a single unit cannot be easily withdrawn from operation; (ii) check the electrical equipment after a power failure, and find that the state of the electrical equipment is inconsistent when it is running alone and when it is tested and checked in the device, such as the operating current, vibration value, and temperature of the equipment, which affects the accuracy of the judgment of the equipment; (iii) affected by the inspection and maintenance cycle, some electrical equipment that is not in continuous operation is judged by employees based on experience during inspection and maintenance; due to the difference in the business ability and experience of each employee, the time selection of maintenance is often not very accurate and reliable, so that there is no good solution to the electrical equipment problems that occur during two adjacent maintenance periods.

2 Literature review

Wang et al. developed an intelligent classification system. The system uses infrared thermal imaging technology to diagnose internal faults of electrical equipment. First, the possible fault areas are manually found through the toolbox of Matlab, and then the red green and blue data of infrared thermal imaging is used as input, and the artificial neural network is used to analyze and obtain the diagnosis result. In the process of experimental analysis, after a certain verification of the selection of the number of neurons in the hidden layer, about 300 infrared images were processed, and the accuracy rate reached 92.82% [3]. Li et al. designed a detection system for detecting basic faults in electrical equipment. The system uses the Zernike moment of the obtained image as its characteristic value and uses support vector machine as a classifier to detect several common errors, such as the damage of the cable joint of the fuse or the failure of the fuse base. The detection system can only detect the most common types of equipment failures with an accuracy of about 83% [4]. Li et al. designed an intelligent thermal imaging diagnostic system to detect the fault of lightning protection system. The system first uses the watershed algorithm to segment the area to be detected and then divides the detection results into three categories: normal, suspicious, and faulty, using a fuzzy neural network. In the specific experimental process, about 100 infrared images were processed, and the accuracy of the system was about 90% [5]. Richoz et al. used a multilayer perceptron neural network for fault detection of electrical equipment. During the experiment, 15 eigenvalues of the infrared image were extracted for the experiment, and 6 eigenvalues were finally selected to be input into the multilayer perceptron network, and the correct rate of the detection results was 79.4% [6]. Amarendra and Reddy developed a real-time offline inspection system based on infrared thermal imaging. The system detects the fault of the electrical system by monitoring the heating area in real time and offline [7]. Mahanta et al. designed a context-dependent fuzzy logic system implemented recursively for hot spot correlation detection of electrical equipment. In the course of the experiment, when it is divided into two categories and three categories according to relevant conditions, the accuracy rates are 92 and 80%, respectively [8]. Pereira et al. proposed a fault diagnosis method based on the finite element analysis of permanent magnet synchronous motor stator winding inter-turn short circuit, established a motor simulation model based on Maxwell 2D, and changed the model parameters to set different short circuit conditions of the motor; the fault degree of the motor is judged according to the influence of the negative sequence current on the load [9]. Kurt and Salamov proposed to establish a motor model based on the finite element analysis in healthy state and fault state and calculate the parameters such as leakage inductance and air gap magnetic flux density of the rotor in the two states; compared with the change of parameters, the model diagnoses whether the fault occurs or not [10]. Kuo and Chang took the inter-turn short-circuit fault of permanent magnet synchronous motor stator winding as the research object, using the spectral characteristics of the fusion vibration signal and the stator current signal to judge the inter-turn short circuit fault, using the improved wavelet packet transform algorithm and fast Fourier transform (FFT), processing and analyzing the acquired signals, thereby effectively diagnosing the existing faults [11]. Using FFT on the signals of motor vibration and stator current, a fractional resampling is proposed to reduce spectral leakage to improve the FFT-based method and to extract and process the characteristic parameters of the fault to diagnose the motor fault [12]. Based on generative adversarial networks and sparse self-encoding deep learning networks, a high-efficiency diagnostic method for winding inter-turn short-circuit faults was given; according to the collected negative sequence current and torque characteristic signals of permanent magnet synchronous motor, the generative adversarial neural network is used to realize the expansion and increase of sample data, and a training set is constructed and combined with a sparse self-encoding network to achieve efficient and accurate fault feature classification and diagnosis [13]. Deng et al. proposed to use a multilayer artificial neural network to diagnose and classify short-circuit faults at different levels in permanent magnet synchronous motor stator windings, through empirical mode decomposition, using neural networks for fault diagnosis [14].

Parlak will use virtual reality (VR) technology; based on the arrangement of the electrical system in the real subway train, the working principle, and the action process, the simulation is carried out, and a set of train electrical simulation system based on VR is designed [15]. This system simulates a subway train realistically, arranges the electrical system in the train body, enhances the three-dimensional sense, facilitates the students to learn and observe, understands the arrangement position of the circuit in the car body, and also facilitates the reference and guidance of the electrical system troubleshooting.

3 Research methods

3.2 Architecture of train electrical simulation system

The main purpose of the train electrical simulation system is for customers, students, and other groups to browse and inspect the electrical system of the train and to learn and study the working principle and action process of the electrical system of the train; in addition, it will also involve the fault simulation and maintenance of the train electrical system [19]. Based on this, the functional architecture of the electrical system obtained from the study is shown in Figure 1.

Figure 1

Functional architecture of train electrical simulation system.

As can be seen from Figure 1, the functional architecture of the system can be explained and analyzed as follows:

1. Display function: The working principle and action process of the train electrical system can be shown to the audience, the branch system to be displayed can be freely selected, and the panoramic display and partial display can be flexibly switched to facilitate the audience to view the details [20].

2. Learning function: Students can freely learn the working principle and structural construction of circuits and other related knowledge and conduct virtual demonstrations. Switch to a branch system that is still weak in understanding and mastery according to its own learning situation and is not disturbed by irrelevant circuit systems.

3. Training function: You can use this system to train new students, break the limitation of time and space, and stay away from the maintenance site; using the simulation of real fault detection and maintenance training, the students can easily master the basic maintenance technology in the virtual training.

4. Assessment function: In the virtual electrical system, the fault point and fault mode are set at will, and it is required to find out the fault point and the cause of the fault, and eliminate the fault, and carry out the fault maintenance assessment [21]. Students can start self-inspection at any time, test their learning effect, and consolidate and digest the learning content.

5. Maintenance function: It can deduce and verify the on-site maintenance plan and maintenance plan and cooperate with the on-site troubleshooting.

In order to realize these functions, control commands are further set in the system, as detailed below.

1. System branch selection (system full display) command: Select the target branch system, and separate the target branch (panoramic view of the overall train/electrical system).

2. Blanking (displaying) command: Blanking (displaying) things other than the target, so that the structure, dynamic relationship, and other information of the doubtful target can be clearly seen.

3. Freeze (unfreeze) command: Indicates that the selected part is faulty (normal) and cannot work normally (can work normally).

In addition, there are commands for scaling, rotation, transparency, etc.

3.3 Functional design of train electrical simulation system

The train electrical system is divided into 15 system branches, such as pantograph lifting system, traction braking system, and door system, so as to facilitate the independent display of a single branch system and eliminate irrelevant line interference [22]. This system is mainly composed of two modules: basic learning and maintenance assessment. Among them, the basic learning module is mainly to learn the basic knowledge of the train electrical system and demonstrate the operation of the circuit and the operation process; in the control demonstration, you can browse the whole panorama as a whole, or display the local system; you can understand the macro-statically; or you can demonstrate the experience dynamically. The maintenance assessment module is an assessment of circuit fault maintenance and trainees’ abilities; it can simulate circuit faults, train employees’ ability to perform fault maintenance, cooperate with on-site maintenance work, conduct troubleshooting drills to guide maintenance personnel to quickly find fault points and eliminate faults, as well as to assess the professional skills of students [23]. The overall structure of the system is shown in Figure 2.

Figure 2

Overall structure of train electrical simulation system.

3.4 Knowledge learning

3.4.1 Basic learning

A display screen is set in the scene of the train electrical simulation system, and the basic knowledge related to the train electrical system is stored in the database in the background of the system; after program control, the display in the scene is associated with this part of the database; by displaying the basic knowledge content, the trainee can learn and master the basic knowledge of the electrical system according to the content on the display and the auxiliary train model [24,25].

3.4.2 Control simulation

After entering the control simulation platform, you can operate and display the electrical system of the train, you can choose to hide or display the car body, or you can make the car body transparent. The animation demonstration process is shown in Figure 3; according to personal needs, select the full display of the electrical system or select the branch system, static circuit, or dynamic circuit; click the action contact; and the demonstration animation starts.

Figure 3

Animation demonstration process diagram.

3.5 Inspection and maintenance

3.5.1 Troubleshooting

The troubleshooting module has two modes: troubleshooting training mode and maintenance plan verification mode. After entering this module, the operation mode can be set first. The troubleshooting training mode is a mode in which students train themselves; students can freely set failure modes, failure points, and failure branch systems; and carry out consolidation training and self-testing on the skills they have learned; The maintenance plan verification mode is a mode used by the fault maintenance unit of the train electrical system to reason and verify its own maintenance plan, use this to reason and demonstrate the maintenance plan before it is officially put into actual maintenance, find out deficiencies and hidden dangers, improve the maintenance plan, and eliminate the possibility of hidden accidents.

3.5.2 Ability assessment

The ability assessment module is an assessment mode for the administrator to examine the students’ maintenance skills; the examiner can freely set the failure mode, failure point, and failure branch system (the system used by the students does not have this authority), and the examiner assess trainees’ mastery of troubleshooting techniques; facing students at different learning stages, the examiner can set the difficulty level of the assessment in a targeted manner.

The troubleshooting process of the electrical system is shown in Figure 4. First, set the fault, then set the detection node freely, and check the working status of the circuit in sections until the fault is eliminated and the circuit can work normally.

Figure 4

Electrical system troubleshooting process.

4 Result analysis

Classify electrical equipment faults, determine the types of electrical equipment faults, and provide a basis for electrical equipment fault diagnosis. Assuming that the historical operation database of electrical equipment is T, the total amount of data is n, and the fault types of electrical equipment are five categories: A, B, C, D, and E are used to identify, respectively, motor failure, inverter failure, abnormal temperature, circuit breaker failure, and power failure. Then, the electrical equipment fault example library is represented as T A , T B , T C , T D , and T E , respectively. Let T A be a positive class; let T B , T C , T D , and T E be a negative class, and the process is expressed as the following formula (1):

where C i belongs to T A and C k belongs to T B , T C , T D , or T E .

Under normal circumstances, the shorter the fault diagnosis time, the better the system performance. In the field of industrial production, the upper limit of fault diagnosis time for electrical equipment is 57 s. The fault diagnosis time results and accuracy results of the design system are obtained through testing, as shown in Table 1 and Figure 5.

Figure 5

Comparison of fault diagnosis accuracy results.

The data in Table 1 and Figure 5 show that in the designed fault diagnosis system, the detection time for various types of faults is shorter than 30 s, whereas the diagnosis time of the manual diagnosis method is 30–52 s. It shows that the electrical equipment fault diagnosis system based on virtual simulation has the advantages of short fault diagnosis time and high efficiency. In addition, the highest diagnostic accuracy of the manual diagnosis method is 75.48%, which is far lower than the accuracy of the diagnostic system. This proves that the designed fault diagnosis system has the advantages of short detection time and high accuracy and can meet the safety requirements of industrial production.

5 Conclusion

The train electrical simulation system truly restores the train electrical system structure; the working principle of the electrical system is realistically simulated, showing the current flowing from the battery and pantograph to the inverter, high-speed circuit breaker and other components, up to the process of the final action parts such as pantograph and traction brake device, demonstrating the action process of the electrical system from the touch switch to the components and action parts; the transformation of the train electrical system from a two-dimensional plan view to a three-dimensional view is realized. In addition, the comparison of the efficiency and accuracy of the designed diagnosis system and the manual diagnosis method is also added; the system has stronger practicability, higher accuracy, and faster diagnosis time.