Partial discharge fault detection method of switchgear based on signal aliasing spectrum separation model

At present, the detection nodes of partial discharge fault in switchgear are generally set in one direction, and the fault detection range is small, which leads to an increase in the false recognition rate of partial discharge fault detection. Therefore, the design and verification analysis of the detection method of partial discharge fault in switchgear is proposed. According to the current measurement requirements and standards, firstly, the characteristics of switchgear discharge faults are extracted, and the fault detection range is expanded by the multi-step method. Multi-step partial discharge fault detection nodes are deployed, and the partial discharge fault detection model with signal aliasing and spectrum separation is constructed. The fault detection is completed by hierarchical fuzzy automatic identification. The test results show that the final detection error rate of partial discharge fault is well controlled below 20% by comparing the seven selected test cycles and combining the signal aliasing spectrum separation model, which shows that the designed detection method of partial discharge fault is more flexible and changeable, and has strong pertinence and reliability. Facing the complex switchgear operating environment, it can also mark the fault position in the shortest time, strengthen the detection results and ensure the detection accuracy.


Introduction
Switchgear is the key equipment for daily power generation, transmission, power dispatching, distribution, electric energy conversion, and elimination control in power systems, which has an important influence on the local safe operation and normal power supply of the power grid [1].However, due to the large coverage and control range of switchgear, arc discharge may occur in local areas of insulation materials, hindering the normal operation of some power grids and reducing overall efficiency [2][3].
To solve this problem, relevant personnel not only regularly inspect, verify and maintain the switchgear, but also design corresponding partial discharge fault detection methods, such as ultrasonic partial discharge detection method, inductive coupling partial discharge detection method, and capacitive partial discharge detection method [4].Although these methods can realize the expected discharge fault detection, they lack pertinence and stability, and it is difficult to capture the best detection point in different background environments [5].
Moreover, the traditional partial discharge fault detection form is generally unidirectional, and the overall efficiency is low, which is also one of the important reasons that affect the final partial discharge fault detection results and cause detection errors [6].Therefore, this paper puts forward the design and verification analysis of the partial discharge fault detection method for switchgear based on the signal aliasing spectrum separation model.The so-called signal aliasing spectrum separation model is a calculation model which reflects the discharge position and fault situation according to signal stacking, spectrum fluctuation, and separation.Overlapping and associating it with the partial discharge fault detection of switchgear can further expand the whole fault detection range, strengthen the detection effect and accuracy from multiple angles, build a more flexible and changeable detection structure, and prevent the spread and deterioration of faults from the root, thus greatly reducing the probability of accidents.

Feature extraction of discharge faults in switchgear.
According to the variation of each band signal, the regular characteristics of the signal are determined and analyzed [7], and the characteristic values of different bands are calculated, as shown in Formula (1): where represents the characteristic values of different bands, represents an identifiable range, represents a stacking area, indicates the frequency of signal fluctuation, and represents a constant value.

Deployment of multistage partial discharge fault detection nodes.
The traditional node arrangement for partial discharge fault detection in switchgear is generally unidirectional, so a directional evaluation structure is established to divide the switchgear into multiple parts [8], and a certain number of fault detection nodes are set up in each region to form a cyclic fault detection structure.
Because the partial discharge detection position of switchgear is changeable, the average distance of fault detection is calculated based on the actual demand, as shown in Formula (2): where indicates the average interval of fault detection, represents an identifiable distance, indicates the identification detection frequency, indicates a detected conversion difference, and represents a duplicate fault detection area [9].Combined with the current measurement, the average interval of fault detection is calculated, and the interval of detection is kept unchanged when detecting partial discharge faults, to improve the overall fault detection results.

Construction of partial discharge fault detection model based on signal aliasing spectrum separation.
We combine the signal aliasing spectrum separation model to extract the characteristics of signal changes during normal operation and discharge fault of switchgear [10], and calculate the fault detection limit, as shown in Formula (3): where indicates the fault detection limit, represents the resolution frequency, indicates the difference of signal fluctuation, indicates the orientation detection range, and represents the detection time domain.Through the model calibration, a base station is set near the node, which is interconnected with the node to expand the separation processing position of the spectrum.The specific model execution structure is shown in Figure 1  According to Figure 1, the design of the model structure of partial discharge fault detection with signal aliasing spectrum separation is completed.

Hierarchical fuzzy automatic recognition.
The so-called hierarchical fuzzy automatic identification mainly refers to the task of fault detection of switchgear through signal overlapping spectrum separation model and hierarchical fault detection structure.First, we set the indicators and parameters of fuzzy automatic identification fault detection.
Then, combined with the signal aliasing spectrum separation model, we make real-time monitoring and analysis of the current switchgear discharge situation, strengthen daily control and management, and adopt the hierarchical fuzzy automatic identification method.Combined with the designed model, we modify and refine the obtained fault detection structure, reduce the fuzzy error, and ensure the reliability of the final detection result.

Test preparation
Combined with the signal aliasing spectrum separation model, the selected switchgear test environment of the H power station is related.The specific experimental environment is shown in Figure 2. Firstly, based on the currently deployed detection nodes, multiple fault detection cycles are set.Each cycle is 48 hours, and a total of 7 cycles are set.Secondly, the basic node data information is collected, summarized, and integrated for subsequent use.We use insulation resistance testers and high-voltage testers, measure the insulation resistance and withstand voltage performance of switchgear to determine the insulation status and quality of the equipment, and ensure that the equipment can withstand the rated voltage under normal operating conditions.Based on this, an oscilloscope is used to observe and record the waveform of electrical signals.During the testing process, we diagnose equipment faults, capture transient signals, and provide measurement data related to voltage, current, frequency, and other aspects.Then, based on this, a multi-dimensional signal overlapping spectrum separation switchgear partial discharge fault detection matrix is constructed, and an intelligent identification program is designed.Next, we calculate the value of the fault detection unit, as shown in Formula (4): where is a value represent a fault detection unit, indicates an identifiable detection area, indicates the detection frequency, represents a controllable detection difference, and indicates the unit detection distance.According to the current measurement requirements, we adjust the fault detection structure at this time, integrate the signal aliasing spectrum separation model, and set the basic test environment and parameters, as shown in Table 1: According to Table 1, the control indicators and parameters of the basic test environment are set and the environment is set.Then, on this basis, combined with the multi-dimensional signal aliasing spectrum separation model, the current fault detection range is delineated, and the nodes are interrelated to form a stable and specific fault identification and detection environment for subsequent testing and verification.

Test process and result analysis
In the above-mentioned test environment, we denoise the signal and calculate the frequency band value of the fault position signal, as shown in Formula (5): where is the frequency band value of the signal representing the fault position, represents a pulse waveform, indicates that fault identification is time-consuming, and indicates the fault correlation area.Combined with the current measurement, the frequency band value of the first fault position signal is calculated, and the abnormal fault position is calibrated twice in the frequency band.Combined with the signal aliasing spectrum separation model, the false recognition rate of partial discharge fault detection is calculated for the set period, as shown in Formula (6):  According to Table 2, the analysis of the test results is completed: according to the comparison of the selected seven test cycles and the signal aliasing spectrum separation model, the final partial discharge fault detection error rate is well controlled below 20%, which shows that the designed partial discharge fault detection method is more flexible and changeable, and has strong pertinence and reliability.Facing the complex switchgear operating environment, it can also mark the fault position and strengthen the detection results in the shortest time.

Conclusion
To sum up, it is the design and verification research of partial discharge fault detection method for switchgear based on signal aliasing spectrum separation model.Compared with the original form of partial discharge fault detection of switchgear, the fault detection structure constructed by combining the signal aliasing spectrum separation model is more flexible and changeable, and it has strong pertinence and stability.In different background environments, it can calibrate the fault bits from multiple angles and extract the abnormal information for subsequent use.In addition, with the aid and support of the signal aliasing spectrum separation model, the currently designed switchgear partial discharge fault detection structure has the effect of collaborative detection.With the fault characteristics as the ultimate guide, the cooperative detection objectives and tasks are formulated, and the "one-to-many" mapping relationship is established.Once the switchgear partial discharge fault occurs, the position can be locked for the first time, and the cross-dislocation identification method is adopted for detection and maintenance, which reduces the probability of related fault problems and avoids causing large-scale power failure accidents.

Figure 1
Figure 1 Structural diagram of partial discharge fault detection model with signal aliasing spectrum separation.
indicates the false recognition rate of partial discharge fault detection, indicates the fault detection range, indicates an orientation detection error, represents the detection average, represents the average number of recognitions, represents the recognition distance, and indicates Table of Control Indicators Parameters of Basic Test Environment.

Table 2 .
International Conference on Energy, Power and Grid(ICEPG 2023)the fault error of the switchgear stack.Combined with the current measurement requirements, the analysis of the test results is completed, as shown in Table 2: Comparative Analysis Table of Test Results. 5