Partial discharge mode identification method, system, equipment and medium

Abstract

The invention provides a partial discharge mode identification method, system and device and a medium, and belongs to the technical field of high voltage and insulation. The method comprises the following steps: simulating a partial discharge phenomenon of the switch cabinet under a laboratory condition, and obtaining a partial discharge pulse waveform by adopting a pulse current method; performing time-domain pulse waveform denoising on the partial discharge pulse waveform by adopting a particle swarm optimization algorithm to generate a partial discharge time-domain pulse waveform, and performing denoising effect inspection; carrying out fast Fourier transform on the time domain pulse waveform of partial discharge to obtain a frequency domain spectrogram, and carrying out time-frequency domain fifteen-dimensional characteristic parameter extraction on kurtosis, asymmetry and contents of different frequency bands according to the time domain waveform and the frequency domain spectrogram of the partial discharge phenomenon; based on a traditional genetic algorithm, a mutation operator and a crossover operator are adjusted, and a target adaptive genetic algorithm is generated; an artificial neural network classifier of a target adaptive genetic algorithm is adopted, and 15-dimensional characteristic parameters are used for training and testing of partial discharge mode recognition.

Description

Technical Field

[0001] The present invention belongs to the technical field of high voltage and insulation, and more specifically relates to a method, system, device and medium for partial discharge pattern recognition. Background Art

[0002] Electrical equipment is a basic component of the power system and the basis for ensuring power supply reliability. Relevant statistical data shows that the deterioration of the insulation performance of electrical equipment is the main cause of its failure, and in severe cases, it may lead to large-scale power outages in the power grid. Partial discharge is not only a sign of insulation deterioration of electrical equipment but also a key factor causing insulation deterioration. The partial discharge pattern recognition of electrical equipment refers to the recognition of the discharge fault types of electrical equipment. For different electrical equipment, the classification of partial discharge fault types is different. The partial discharge fault types of transformers mainly include surface discharge, air gap discharge and corona discharge, while gas insulated switchgear also includes two types: floating discharge and free metal discharge. The fault type is closely related to the severity of partial discharge. Therefore, the effective recognition of the electrical equipment fault type can provide reliable reference information for the assessment of the partial discharge risk level.

[0003] At present, for partial discharge pattern recognition, relevant technical personnel have conducted a large number of experimental studies and achieved remarkable results. The selection of the classifier is one of the key links in pattern recognition. Commonly used classifiers include: Bayesian classifier, neural network, support vector machine, etc. Technical personnel have optimized and improved their classification algorithms to overcome the deficiencies of common classifiers and improve the partial discharge pattern recognition rate. Common optimization algorithms include: particle swarm optimization algorithm, grey wolf optimization algorithm, genetic algorithm, etc.

[0004] As a global search algorithm, the traditional genetic algorithm has advantages such as parallelism and self-organization, and can better solve complex optimization problems. However, the selection of its parameters, especially the crossover probability Pc and mutation probability Pm, has a crucial impact on the performance and results of the algorithm. In the evolution process of the traditional genetic algorithm, these parameters are determined in advance, which often leads to the algorithm being prone to falling into a local optimal solution in the later stage of evolution, that is, the so-called "premature" phenomenon. Once the "premature" phenomenon occurs, the algorithm will be difficult to jump out of the current search area and thus unable to find the global optimal solution.

[0005] In addition, the convergence speed of the genetic algorithm is also an issue that cannot be ignored. In the initial stage of evolution, due to the large differences among individuals in the population, the algorithm needs to spend a lot of time searching and screening; while in the later stage of evolution, as the differences among individuals in the population decrease, the convergence speed of the algorithm will gradually slow down and may even stagnate. This imbalance in the convergence speed not only affects the efficiency of the algorithm but also limits its application in scenarios with high requirements for real-time and online performance.

[0006] Therefore, although optimization algorithms such as genetic algorithms have achieved remarkable results in fields such as partial discharge pattern recognition, their defects and limitations cannot be ignored. Summary of the Invention

[0007] In view of the above problems, the purpose of the present invention is to provide a partial discharge pattern recognition method, system, device and medium, which uses an improved artificial neural network optimized by an adaptive genetic algorithm based on dominant inheritance for partial discharge pattern recognition, effectively improving the accuracy of partial discharge pattern recognition.

[0008] To achieve the above object, the present invention is realized through the following technical solutions: In a first aspect, an embodiment of the present application provides a partial discharge pattern recognition method, including: Simulate the partial discharge phenomenon of the switch cabinet under laboratory conditions, and obtain the partial discharge pulse waveform by using the pulse current method; wherein, the partial discharge phenomenon includes corona, surface discharge, and air gap discharge phenomena; Use the particle swarm optimization algorithm to denoise the time-domain pulse waveform of the partial discharge, generate the time-domain pulse waveform of the partial discharge, and perform denoising effect verification; Perform a fast Fourier transform on the time-domain pulse waveform of the partial discharge to obtain the frequency-domain spectrogram, extract the fifteen-dimensional characteristic parameters in the time-frequency domain for the time-domain waveform and frequency-domain spectrogram of the partial discharge phenomenon, and generate the fifteen-dimensional characteristic parameters; Based on the traditional genetic algorithm, adjust the mutation operator and the crossover operator to generate the target adaptive genetic algorithm; Use the artificial neural network classifier of the target adaptive genetic algorithm to perform training and testing on partial discharge pattern recognition by using the fifteen-dimensional characteristic parameters.

[0009] In an alternative embodiment, the step of simulating the partial discharge phenomenon of the switch cabinet under laboratory conditions and obtaining the partial discharge pulse waveform by using the pulse current method includes: Build a switch cabinet model with needle-plate, surface, and internal defects under laboratory conditions to simulate the corona, surface discharge, and air gap discharge phenomena of the switch cabinet; Use the pulse current method to detect the switch cabinet model, pressurize the switch cabinet model by the constant voltage method, increase the effective value of the AC voltage source from 0 at a rate of 1 kV per minute to 1.2 times the partial discharge inception voltage and then stop, collect a set of partial discharge data every 20 minutes, and the sampling time lasts for 5 minutes to obtain the partial discharge pulse waveform.

[0010] In an alternative embodiment, the step of using the particle swarm optimization algorithm to denoise the time-domain pulse waveform of the partial discharge, generate the time-domain pulse waveform of the partial discharge, and perform denoising effect verification includes: The multi - layer decomposition of the partial discharge pulse waveform signal is carried out by using the orthogonal discrete wavelet transform to obtain the wavelet coefficients of each layer; For the high - frequency wavelet coefficients of each layer, soft thresholding is used for threshold quantization, and the steepest descent method is used for iterative calculation of the wavelet threshold to complete the denoising of the time - domain pulse waveform and generate the time - domain pulse waveform of partial discharge; Based on the time - domain pulse waveform of partial discharge, the partial discharge waveform is simulated by simulating the exponential decay waveform and the exponential oscillation waveform, and white noise and narrow - band interference are added to simulate noise for testing the denoising effect.

[0011] In an alternative embodiment, the iterative calculation of the wavelet threshold using the steepest descent method includes: Iteratively calculate the inertia weight w using a linear change strategy;

[0012] where w max is the maximum inertia weight, w min is the minimum inertia weight, iter is the current iteration number, and iter max is the maximum iteration number.

[0013] In an alternative embodiment, the fast Fourier transform of the time - domain pulse waveform of partial discharge is performed to obtain the frequency - domain spectrogram, and fifteen - dimensional characteristic parameters in the time - frequency domain are extracted from the time - domain waveform and the frequency - domain spectrogram of the partial discharge phenomenon to generate fifteen - dimensional characteristic parameters, including: The fast Fourier transform of the time - domain pulse waveform of partial discharge is performed to obtain the frequency - domain spectrogram, the intensity of different frequency components in the frequency - domain spectrogram is analyzed, and the key frequency components are determined; Based on the key frequency components, waveform features, frequency - domain features, and time - frequency domain features are extracted to obtain fifteen - dimensional characteristic parameters; Based on the fifteen - dimensional characteristic parameter fingerprint parameter map, the logarithm of the ordinate is processed; The fifteen - dimensional characteristic parameters include: kurtosis, skewness, rise - time, fall - time, 50% amplitude pulse duration, 10% amplitude pulse duration, asymmetry, pulse factor, margin factor, maximum discharge amplitude, maximum amplitude frequency, 0 - 30Hz average amplitude, 30 - 50Hz average amplitude, equivalent time width, and equivalent frequency width.

[0014] In an alternative embodiment, based on the traditional genetic algorithm, the mutation operator and the crossover operator are adjusted to generate the target adaptive genetic algorithm, including: Based on the traditional genetic algorithm, the adaptive crossover probability P c and the adaptive mutation probability P m are calculated through the following formula:

[0015] Among them, Pc1 is the crossover probability of individuals with fitness higher than the average fitness, Pc2 is the crossover probability of individuals with fitness of 0, Pm1 is the mutation probability of individuals with fitness lower than the average fitness, Pm2 is the mutation probability of individuals with fitness of 0, f is the fitness of the mutated individual, fmax is the maximum fitness of the population, and f′ is the larger fitness value among the two crossed individuals. is the average fitness.

[0016] In an optional embodiment, the artificial neural network classifier using the target adaptive genetic algorithm performs training and testing for partial discharge pattern recognition using fifteen-dimensional feature parameters, including: Obtain the artificial neural network classifier of the target adaptive genetic algorithm, set the number of input layer nodes to 15, the number of output layer nodes to 1, the number of hidden layer nodes to 31, the learning rate to 0.01, allow the maximum deviation to be 0.00001, and the maximum number of iterations to be 1000 times; Extract 130 groups of fifteen-dimensional feature parameters, use 100 groups of fifteen-dimensional feature parameters as the training set and 30 groups of fifteen-dimensional feature parameters as the test set to train and optimize the artificial neural network classifier; Obtain the fifteen-dimensional feature parameters of the switch cabinet in real time, and use the optimized artificial neural network classifier to generate the recognition result of the partial discharge pattern.

[0017] In a second aspect, the embodiment of the present application further provides a partial discharge pattern recognition system, including: A data acquisition module, which is used to simulate the partial discharge phenomenon of the switch cabinet under laboratory conditions and obtain the partial discharge pulse waveform by using the pulse current method; among them, the partial discharge phenomenon includes corona, surface discharge, and air gap discharge phenomena; A denoising module, which is used to perform time-domain pulse waveform denoising on the partial discharge pulse waveform by using the particle swarm optimization algorithm, generate the time-domain pulse waveform of the partial discharge, and perform denoising effect inspection; A feature extraction module, which is used to perform fast Fourier transform on the time-domain pulse waveform of the partial discharge to obtain the frequency-domain spectrogram, and extract the fifteen-dimensional feature parameters in the time-frequency domain for the time-domain waveform and frequency-domain spectrogram of the partial discharge phenomenon to generate the fifteen-dimensional feature parameters; An algorithm construction module, which is used to adjust the mutation operator and crossover operator based on the traditional genetic algorithm to generate the target adaptive genetic algorithm; A pattern recognition module, which is used to use the artificial neural network classifier of the target adaptive genetic algorithm to perform training and testing for partial discharge pattern recognition using fifteen-dimensional feature parameters.

[0018] In a third aspect, an embodiment of the present application further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the partial discharge pattern recognition method described in any one of the above are implemented.

[0019] In a fourth aspect, an embodiment of the present application further provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the partial discharge pattern recognition method described in any one of the above are implemented.

[0020] As can be seen from the above technical solutions, the present invention has the following advantages: In the partial discharge pattern recognition method provided by the present application, through equivalent simulation experiments on common partial discharge defects of common high-voltage electrical equipment under laboratory conditions, a partial discharge sample set is obtained. Through a series of preprocessing such as denoising, and feature parameter extraction is performed on its time-domain and frequency-domain information. An improved artificial neural network optimized by an adaptive genetic algorithm based on dominant inheritance is used for partial discharge pattern recognition, effectively improving the accuracy of partial discharge pattern recognition.

[0021] The present application optimizes the fixed crossover and mutation parameters into a dynamic adaptive genetic algorithm adjusted according to fitness, solves the premature problem of the traditional genetic algorithm, can effectively improve the performance of global optimization, and has very good robustness.

[0022] When preprocessing the waveform, the present application uses wavelet denoising of the particle swarm optimization algorithm, which can effectively retain the characteristic information of partial discharge while filtering out white noise and narrowband interference, greatly improving the accuracy of subsequent partial discharge pattern recognition.

[0023] The method proposed by the present application is applicable to various specifications of power equipment under various voltage levels and has strong portability. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0025] Figure 1 It is a schematic flow chart of the partial discharge pattern recognition method provided by the present application.

[0026] Figure 2 It is a schematic structural diagram of the experimental circuit provided by the present application.

[0027] Figure 3Schematic flow chart of the wavelet denoising method based on the particle swarm optimization algorithm provided by this application.

[0028] Figure 4 Schematic structural diagram of the partial discharge pattern recognition system provided by this application.

[0029] Figure 5 Schematic structural diagram of the electronic device provided by this application. Detailed implementation manners

[0030] In the following, various embodiments of the present disclosure will be described more fully in the specific steps of the partial discharge pattern recognition method to be described in detail below. The present disclosure can have various embodiments and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents and / or alternative solutions falling within the spirit and scope of the various embodiments of the present disclosure.

[0031] Hereinafter, the term "include" or "may include" that can be used in various embodiments of the present disclosure indicates the presence of the disclosed functions, operations or elements, and does not limit the addition of one or more functions, operations or elements. In addition, as used in various embodiments of the present disclosure, the terms "include", "have" and their cognates are only intended to indicate specific features, numbers, steps, operations, elements, components or combinations of the foregoing items, and should not be understood as first excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing items or the possibility of adding one or more features, numbers, steps, operations, elements, components or combinations of the foregoing items.

[0032] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0033] Please refer to Figure 1 The following is a method flow chart of a partial discharge pattern recognition method in a specific embodiment. The method includes: S1: Simulate the partial discharge phenomenon of the switch cabinet under laboratory conditions, and obtain the partial discharge pulse waveform by using the pulse current method; wherein, the partial discharge phenomenon includes corona, surface discharge and air gap discharge phenomena.

[0034] In the specific implementation manner, taking the partial discharge detection of a 35 kV switchgear as an example, a switchgear model including needle-plate, surface, and internal defects was constructed under laboratory conditions to simulate the corona, surface, and air-gap discharges of the switchgear, and the partial discharge pulse waveform was obtained by the pulse current method.

[0035] Among them, data acquisition can be realized by building an experimental circuit. To reduce partial discharge in the experimental circuit, copper was used to wrap the wire connection points and the grounding points, and the bending of the connecting wires of the experimental equipment was minimized as much as possible. As Figure 2 shown, the construction of this experimental circuit complies with the IC 60270 standard. Through the AC withstand voltage test, the corona inception voltage of this experimental circuit is high and will not affect the subsequent partial discharge detection, meeting the experimental requirements.

[0036] The partial discharge detection method selected for the experiment is the pulse current method, which has become the most widely used and technically mature method in this field. The voltage application method is the constant voltage method, which is more in line with the step-free voltage condition faced in actual operation than the step-up voltage method. The set voltage is 1.2 times the partial discharge inception voltage corresponding to the model. When measuring the partial discharge inception voltage of a certain defect, the effective value of the AC applied voltage starts from 0 and rises at a rate of 5 minutes per kilovolt until stable partial discharge occurs in the switchgear model. At this time, the effective value of the AC voltage source is the partial discharge inception voltage. In the formal experiment of measuring the partial discharge waveform and spectrogram, the effective value of the AC voltage source starts from 0 and is increased to 1.2 times the partial discharge inception voltage at a rate of 1 minute per kilovolt and then stops. A set of partial discharge data is collected every 20 minutes, the sampling time lasts for 5 minutes, and 100 typical discharge waveforms are collected simultaneously.

[0037] S2: The particle swarm optimization algorithm is used to denoise the partial discharge pulse waveform in the time domain, generate the time-domain pulse waveform of the partial discharge, and conduct a denoising effect test.

[0038] In the specific implementation manner, first, the orthogonal discrete wavelet transform is used to decompose the partial discharge pulse waveform signal into multiple layers to obtain the wavelet coefficients of each layer. The soft threshold is used for threshold quantization of the high-frequency wavelet coefficients of each layer, and the steepest descent method is used for iterative calculation of the wavelet threshold.

[0039] Among them, the change of the inertia weight adopts a linear change strategy, that is, as the number of iterations increases, the inertia weight continuously decreases, which can enable the particle swarm to have strong global convergence ability in the initial stage of the algorithm and strong local convergence ability in the later stage of the algorithm.

[0040] Specifically, the calculation formula of the inertia weight w is as follows:

[0041] where wmax is the maximum inertia weight, w min is the minimum inertia weight, iter is the current iteration number, iter max is the maximum iteration number.

[0042] At this time, in order to verify the denoising effect of the wavelet denoising method optimized by the particle swarm algorithm, an exponentially decaying waveform and an exponentially oscillating waveform were simulated to simulate the partial discharge waveform, and white noise and narrowband interference were added to simulate the noise. The sampling rate was taken as 109, the amplitude of the exponentially decaying waveform was 0.01 V, and the time constant was 2.5×10 -6 S; the amplitude of the exponentially oscillating waveform was 0.03 V, and the time constant was 5×10 -6 S; the amplitude of the white noise was 0.001 V, the amplitude of the narrowband interference was 0.001 V, and the frequencies were 105, 3.5×105, 5.5×105, and 10×105 Hz. Good denoising effects were achieved.

[0043] As an example, in this step, a wavelet denoising method based on the particle swarm optimization algorithm is disclosed, as Figure 3 shown, the specific process is as follows: 1. Perform wavelet decomposition on the partial discharge signal to obtain wavelet decomposition coefficients at each scale.

[0044] 2. Input the initial conditions, generate the initial population, and initialize the particle velocity.

[0045] 3. Calculate the initial fitness, and update the initial individual, global optimal position, and optimal objective.

[0046] 4. Determine whether the preset maximum iteration number is reached. If so, execute step 5; if not, execute step 7.

[0047] 5. Obtain the optimal wavelet threshold and select the optimal particle as the optimal individual.

[0048] 6. Perform threshold processing on each scale wavelet, and use the inverse discrete wavelet transform to obtain the denoised partial discharge pulse waveform signal.

[0049] 7. Continuously iterate to update the positions and velocities of the particles.

[0050] 8. Calculate the fitness value of each particle.

[0051] 9. Update the individual historical optimal fitness value and position of each particle.

[0052] 10. Update the group historical optimal fitness value and position.

[0053] 11. Update the inertia weight, iteration number, and execute step 5.

[0054] S3: Perform a fast Fourier transform on the time-domain pulse waveform of partial discharge to obtain a frequency-domain spectrogram, extract fifteen-dimensional characteristic parameters in the time-frequency domain for the time-domain waveform and frequency-domain spectrogram of the partial discharge phenomenon, and generate fifteen-dimensional characteristic parameters.

[0055] In the specific implementation manner, perform a fast Fourier transform (FFT) on the time-domain pulse waveform of partial discharge to obtain a frequency-domain spectrogram, and extract fifteen-dimensional characteristic parameters in the time-frequency domain for the time-domain waveforms and frequency-domain spectrograms of three types of partial discharge in terms of kurtosis, skewness, and content in different frequency bands.

[0056] Specifically, for the three types of partial discharge, use the platform and test method built in step one to extract waveforms and spectra, then calculate 3 groups of fingerprint feature quantities respectively and conduct a visual comparison. Since the orders of magnitude of the 15 fingerprint parameters vary greatly, the logarithm of the ordinate is taken when drawing the fingerprint parameter diagram. In addition, for corona discharge and surface discharge, the skewness of the characteristic quantity is negative, so the absolute value is taken. It can be found that the fingerprint parameters of the same type of partial discharge are almost consistent and stable, while the fingerprint parameters of different types of partial discharge, although similar in some characteristic quantities in part, have large differences overall, so they can be used as the basis for partial discharge pattern recognition.

[0057] Among them, the extracted fifteen-dimensional characteristic parameters are specifically shown in Table 1 below: Table 1: Comparison table of characteristic parameters

[0058] S4: Based on the traditional genetic algorithm, adjust the mutation operator and crossover operator to generate a target adaptive genetic algorithm.

[0059] In the specific implementation manner, the purpose of this step is to change the situation where the coefficients of crossover and mutation in the traditional genetic algorithm are artificially set initially. When designing the mutation operator, make individuals with low fitness mutate with a higher probability, and when designing the crossover operator, make individuals with high fitness cross with a higher probability. Construct a new adaptive genetic algorithm (abbreviation: P-AGA).

[0060] Specifically, when designing the crossover operator, make individuals with high fitness cross with a higher probability because after the crossover operation, the possibility of inheriting dominant genes to the offspring is greater, which is more in line with the law of natural genetic evolution; when designing the mutation operator, make individuals with low fitness mutate with a higher probability, which can mutate out more dominant patterns. At the same time, reasonable crossover and mutation can more effectively produce dominant individuals and jump out of the local optimum. Therefore, a new adaptive genetic algorithm is designed, which can effectively solve the premature phenomenon of the algorithm to a certain extent. To ensure that the best individuals in each generation are not destroyed, an elitist selection strategy is adopted to ensure that the algorithm finally converges to the global optimum adaptive crossover probability P cand the adaptive mutation probability P m , make the following adjustments:

[0061] Among them, Pc1 is the crossover probability of individuals with fitness higher than the average fitness, Pc2 is the crossover probability of individuals with fitness of 0, Pm1 is the mutation probability of individuals with fitness lower than the average fitness, Pm2 is the mutation probability of individuals with fitness of 0, f is the fitness of the mutated individual, fmax is the maximum fitness of the population, and f′ is the larger fitness value among the two crossed individuals. is the average fitness. According to experiments, the coefficients in the formula are taken as Pc1 = 0.9, Pc2 = 0.6, Pm1 = 0.5, and Pm2 = 0.5.

[0062] S5: Use the artificial neural network classifier of the target adaptive genetic algorithm to perform training and testing on partial discharge pattern recognition using fifteen-dimensional feature parameters.

[0063] In the specific implementation, the artificial neural network classifier optimized by the improved genetic algorithm uses the fifteen-dimensional feature parameters extracted in step S3 to perform training and testing on partial discharge pattern recognition.

[0064] First, obtain the artificial neural network classifier of the target adaptive genetic algorithm. Among them, the main parameters of the artificial neural network are the number of nodes in the input layer, hidden layer, and output layer. The number of nodes in the input layer and output layer is determined by the actual problem. The determination of the number of nodes in the hidden layer will greatly affect the number of iterations and classification accuracy. According to experience, the number of nodes in the hidden layer is determined by 2n + 1, where n is the number of nodes in the input layer. The number of nodes in the input layer of this step is determined according to the dimension of the feature quantity, so 15 is taken, and the output is the recognition type, so 1 is taken for the output layer, and 31 is taken for the hidden layer according to experience. In addition, there are also the learning rate, the maximum allowable deviation, and the maximum number of iterations. The learning rate affects the update amplitude of the iteration. When the learning rate is large, it may lead to faster convergence, but it is also easy to fall into the local minimum point; while when the learning rate is small, it will greatly extend the iteration time and increase the number of iterations, affecting the efficiency. The maximum allowable deviation and the maximum number of iterations are important factors affecting the termination of the iteration. When the actual deviation is less than the maximum allowable deviation or the number of iterations reaches the maximum number of iteration limits, the iteration exits. Therefore, in this step, the learning rate is selected as 0.01, the maximum allowable deviation is 0.00001, and the maximum number of iterations is 1000 times.

[0065] Then, 130 sets of discharge data for each of corona discharge, surface discharge, and internal discharge are extracted. Among them, 100 sets of each type of partial discharge are used as the training set, and 30 sets are used as the test set. A total of 15-dimensional feature quantities are taken in the way of constructing the partial discharge pattern recognition fingerprint library in Step 3, and the improved genetic algorithm is used to optimize the initial values of the artificial neural network. The recognition accuracy of each category in the training set reaches 98%, 100%, and 100%, and the overall recognition accuracy reaches 99%. In the test set, only three sets of surface discharges are misrecognized as internal discharges. The recognition rate of corona discharge reaches 100%, and the overall recognition accuracy reaches 93.33%, achieving good recognition results.

[0066] In this embodiment, through laboratory simulation and the pulse current method, the discharge waveform is obtained. Combining the particle swarm optimization algorithm for denoising and time-frequency domain feature extraction, and using the adjusted target adaptive genetic algorithm to optimize the artificial neural network classifier, the efficient and accurate recognition of corona, surface, and air gap discharge phenomena is realized, and the accuracy and reliability of partial discharge pattern recognition are improved.

[0067] As Figure 4 shown, the following is an embodiment of the partial discharge pattern recognition system provided by the present disclosure. This system and the partial discharge pattern recognition methods of the above embodiments belong to the same inventive concept. For the details not described in detail in the embodiment of the partial discharge pattern recognition system, reference can be made to the embodiments of the above partial discharge pattern recognition methods.

[0068] A partial discharge pattern recognition system includes: a data acquisition module 1, a denoising module 2, a feature extraction module 3, an algorithm construction module 4, and a pattern recognition module 5.

[0069] The data acquisition module 1 is used to simulate the partial discharge phenomenon of the switch cabinet under laboratory conditions and obtain the partial discharge pulse waveform by the pulse current method; wherein, the partial discharge phenomenon includes corona, surface, and air gap discharge phenomena.

[0070] The denoising module 2 is used to denoise the time-domain pulse waveform of the partial discharge by using the particle swarm optimization algorithm, generate the time-domain pulse waveform of the partial discharge, and perform denoising effect inspection.

[0071] The feature extraction module 3 is used to perform fast Fourier transform on the time-domain pulse waveform of the partial discharge to obtain the frequency-domain spectrogram, extract the fifteen-dimensional feature parameters in the time-frequency domain for the time-domain waveform and frequency-domain spectrogram of the partial discharge phenomenon, and generate the fifteen-dimensional feature parameters.

[0072] The algorithm construction module 4 is used to adjust the mutation operator and crossover operator based on the traditional genetic algorithm to generate the target adaptive genetic algorithm.

[0073] The pattern recognition module 5 is used to train and test the partial discharge pattern recognition by using an artificial neural network classifier with a target adaptive genetic algorithm and fifteen-dimensional feature parameters.

[0074] The partial discharge pattern recognition system provided in this embodiment simulates the partial discharge phenomenon of the switch cabinet, collects the pulse waveforms, denoises them by using the particle swarm optimization algorithm, extracts the time-frequency domain feature parameters, and trains the artificial neural network classifier by using the optimized adaptive genetic algorithm, realizing the efficient and accurate recognition of the partial discharge pattern and improving the stability and reliability of the recognition.

[0075] Figure 5 It is a schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present invention.

[0076] The partial discharge pattern recognition method provided in the embodiments of this application can be applied to an electronic device. Those skilled in the art can understand that the structure of the electronic device involved in the embodiments of the present invention does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements. In the embodiments of the present invention, the electronic device includes, but is not limited to, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown in the figure, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the embodiments of the present application described and / or claimed herein.

[0077] The electronic device may include a processor, an external memory interface, an internal memory, a universal serial bus (USB) interface, a charging management module, a power management module, a battery, a wireless communication module, an audio module, a speaker, a microphone, a sensor module, a key, a camera, a display screen, and a SIM card interface, etc.

[0078] The processor may include one or more processing units. For example, the processor may include a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Among them, different processing units may be independent devices or integrated in one or more processors.

[0079] Among them, the processor may be the nerve center and command center of the electronic device. The controller may generate operation control signals according to the instruction operation code and timing signals to complete the control of fetching and executing instructions.

[0080] A memory may also be provided in the processor for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. This memory can store the instructions or data that the processor has just used or recycled. If the processor needs to use the instruction or data again, it can directly call it from this memory. This avoids repeated accesses, reduces the waiting time of the processor, and thus improves the system efficiency.

[0081] The external memory interface can be used to connect an external memory card, such as a MicroSD card, to expand the storage capacity of the electronic device. The external memory card communicates with the processor through the external memory interface to achieve the data storage function. For example, files such as music and videos are saved in the external memory card.

[0082] The internal memory can be used to store computer-executable program code, and the computer-executable program code includes instructions. The processor executes various functional applications and data processing of the electronic device by running the instructions stored in the internal memory. The internal memory may include a program storage area and a data storage area. The internal memory may include a high-speed random access memory and may also include non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, a universal flash storage (UFS), etc.

[0083] The wireless communication function of the electronic device can be implemented through an antenna, a wireless communication module, a modem processor, a baseband processor, etc.

[0084] The wireless communication module can provide solutions for wireless communications applied to electronic devices, including wireless local area networks (WLANs) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite systems (GNSS), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc.

[0085] The electronic device can implement audio functions through an audio module, a speaker, a receiver, a microphone, a headphone jack, and an application processor, etc.

[0086] The electronic device can implement a shooting function through an ISP, a camera, a video codec, a GPU, a display screen, and an application processor, etc.

[0087] The electronic device can implement a display function through a GPU, a display screen, and an application processor, etc.

[0088] The GPU is a microprocessor for image processing, connecting the display screen and the application processor. The GPU is used to perform mathematical and geometric calculations for graphics rendering. The processor may include one or more GPUs, which execute program instructions to generate or change display information.

[0089] The display screen is used to display images, videos, etc. The display screen includes a display panel.

[0090] The above-mentioned electronic device implements the local discharge pattern recognition method of the present application. By comprehensively applying advanced technologies such as simulation experiments, waveform denoising, feature extraction, and optimization algorithms, the recognition accuracy and efficiency of the local discharge phenomenon are significantly enhanced, providing strong support for the condition monitoring and health management of electrical equipment such as switch cabinets, and achieving the beneficial effects of the intelligence and automation of the recognition process.

[0091] In the storage medium provided by the present application, there is a program product capable of implementing the local discharge pattern recognition method.

[0092] The partial discharge pattern recognition method includes: simulating the partial discharge phenomenon of the switch cabinet under laboratory conditions, and obtaining the partial discharge pulse waveform by using the pulse current method; wherein, the partial discharge phenomenon includes corona, surface discharge, and air gap discharge phenomena; using the particle swarm optimization algorithm to denoise the partial discharge pulse waveform in the time domain, generating the time-domain pulse waveform of the partial discharge, and performing denoising effect inspection; performing fast Fourier transform on the time-domain pulse waveform of the partial discharge to obtain the frequency-domain spectrogram, and extracting the fifteen-dimensional characteristic parameters in the time-frequency domain for the kurtosis, asymmetry degree, and content of different frequency bands according to the time-domain waveform and frequency-domain spectrogram of the partial discharge phenomenon, generating the fifteen-dimensional characteristic parameters; based on the traditional genetic algorithm, adjusting the mutation operator and crossover operator to generate the target adaptive genetic algorithm; using the artificial neural network classifier of the target adaptive genetic algorithm to perform training and testing for partial discharge pattern recognition by using the fifteen-dimensional characteristic parameters.

[0093] In some possible implementation manners, the partial discharge pattern recognition method of the present disclosure can be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of this specification.

[0094] The storage medium of the present disclosure may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0095] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for identifying a partial discharge pattern, characterized in that: include: The partial discharge phenomenon of the switch cabinet is simulated under laboratory conditions, and the pulse current method is used to obtain the partial discharge pulse waveform; the partial discharge phenomenon includes corona, surface discharge, and air gap discharge; The particle swarm optimization algorithm is used to denoise the time domain pulse waveform of the partial discharge pulse waveform, generate the time domain pulse waveform of the partial discharge, and test the denoising effect; Performing fast Fourier transform on the time-domain pulse waveform of partial discharge to obtain a frequency-domain spectrum, extracting fifteen-dimensional characteristic parameters in the time-frequency domain from the time-domain waveform and the frequency-domain spectrum of the partial discharge phenomenon, and generating fifteen-dimensional characteristic parameters; Based on the traditional genetic algorithm, the mutation operator and crossover operator are adjusted to generate a target adaptive genetic algorithm; An artificial neural network classifier based on target adaptive genetic algorithm is used to train and test partial discharge pattern recognition using fifteen-dimensional feature parameters.

2. The partial discharge pattern recognition method according to claim 1, characterized in that: The method of simulating the partial discharge phenomenon of the switch cabinet under laboratory conditions and obtaining the partial discharge pulse waveform by using the pulse current method includes: By building a switch cabinet model with pin plate, surface and internal defects under laboratory conditions, the corona, surface and air gap discharge phenomena of the switch cabinet are simulated; The switch cabinet model was tested using the pulse current method, and the switch cabinet model was pressurized using the constant voltage method. The effective value of the AC voltage source was increased from 0 at a preset rate to 1.2 times the partial discharge starting voltage and then stopped. A set of partial discharge data was collected every 20 minutes, and the sampling time lasted for 5 minutes to obtain the partial discharge pulse waveform.

3. The partial discharge pattern recognition method according to claim 1, characterized in that: The method of using a particle swarm optimization algorithm to perform time domain pulse waveform denoising on a partial discharge pulse waveform, generating a time domain pulse waveform of a partial discharge, and performing a denoising effect test includes: Orthogonal discrete wavelet transform is used to decompose the partial discharge pulse waveform signal into multiple layers to obtain the wavelet coefficients of each layer. The soft threshold is used to quantize the high-frequency wavelet coefficients of each layer, and the steepest descent method is used to iteratively calculate the wavelet threshold to complete the time domain pulse waveform denoising and generate the time domain pulse waveform of partial discharge; Based on the time-domain pulse waveform of partial discharge, the partial discharge waveform is simulated by simulating the exponential decay waveform and the exponential oscillation waveform, and white noise and narrow-band interference simulation noise are added to test the denoising effect.

4. The partial discharge pattern recognition method according to claim 3, characterized in that: The iterative calculation of the wavelet threshold using the steepest descent method includes: The inertia weight w is iteratively calculated using a linear change strategy; Among them, w max is the maximum inertia weight, w min is the minimum inertia weight, iter is the current iteration number, iter max is the maximum number of iterations.

5. The partial discharge pattern recognition method according to claim 1, characterized in that: The method of performing fast Fourier transform on the time domain pulse waveform of the partial discharge to obtain a frequency domain spectrogram, extracting fifteen-dimensional characteristic parameters in the time and frequency domains from the time domain waveform and the frequency domain spectrogram of the partial discharge phenomenon, and generating fifteen-dimensional characteristic parameters includes: Perform fast Fourier transform on the time-domain pulse waveform of partial discharge to obtain a frequency-domain spectrum, analyze the intensity of different frequency components in the frequency-domain spectrum, and determine the key frequency components; Extract waveform features, frequency domain features and time-frequency domain features based on key frequency components to obtain fifteen-dimensional feature parameters; Based on the 15-dimensional characteristic parameter fingerprint parameter map, the vertical coordinate is logarithmically processed; The fifteen-dimensional characteristic parameters include: kurtosis, skewness, rising edge time, falling edge time, 50% amplitude pulse duration, 10% amplitude pulse duration, asymmetry, pulse factor, margin factor, maximum discharge amplitude, maximum amplitude frequency, 0-30Hz average amplitude, 30-50Hz average amplitude, equivalent time width and equivalent frequency bandwidth.

6. The partial discharge pattern recognition method according to claim 1, characterized in that: The method of adjusting the mutation operator and the crossover operator based on the traditional genetic algorithm to generate a target adaptive genetic algorithm includes: Based on the traditional genetic algorithm, the adaptive crossover probability P is calculated by the following formula c and adaptive mutation probability P m : Among them, Pc1 is the crossover probability of an individual with a fitness higher than the average fitness, Pc2 is the crossover probability of an individual with a fitness of 0, Pm1 is the mutation probability of an individual with a fitness lower than the average fitness, Pm2 is the mutation probability of an individual with a fitness of 0, f is the fitness of the mutated individual, fmax is the maximum fitness of the population, and f′ is the larger fitness value of the two crossover individuals. is the average fitness value.

7. The partial discharge pattern recognition method according to claim 6, characterized in that: The artificial neural network classifier using the target adaptive genetic algorithm uses fifteen-dimensional characteristic parameters to perform training and testing for partial discharge pattern recognition, including: Obtain the artificial neural network classifier of the target adaptive genetic algorithm, set the number of input layer nodes to 15, the number of output layer nodes to 1, the number of hidden layer nodes to 31, the learning rate to 0.01, the maximum allowed deviation to 0.00001, and the maximum number of iterations to 1000; 130 groups of 15-dimensional feature parameters were extracted, 100 groups of 15-dimensional feature parameters were used as training sets, and 30 groups of 15-dimensional feature parameters were used as test sets to train and optimize the artificial neural network classifier; The fifteen-dimensional characteristic parameters of the switch cabinet are obtained in real time, and the recognition results of the partial discharge mode are generated using the optimized artificial neural network classifier.

8. A partial discharge pattern recognition system, characterized in that: The system adopts the partial discharge pattern recognition method as claimed in any one of claims 1 to 7; The system comprises: The data acquisition module is used to simulate the partial discharge phenomenon of the switch cabinet under laboratory conditions and obtain the partial discharge pulse waveform using the pulse current method; the partial discharge phenomenon includes corona, surface discharge, and air gap discharge; A denoising module is used to denoise the time-domain pulse waveform of the partial discharge pulse waveform using a particle swarm optimization algorithm, generate the time-domain pulse waveform of the partial discharge, and perform denoising effect testing; The feature extraction module is used to perform fast Fourier transform on the time domain pulse waveform of partial discharge to obtain a frequency domain spectrum, extract fifteen-dimensional feature parameters in the time and frequency domains from the time domain waveform and frequency domain spectrum of the partial discharge phenomenon, and generate fifteen-dimensional feature parameters; Algorithm building module, used to adjust mutation operator and crossover operator based on traditional genetic algorithm to generate target adaptive genetic algorithm; The pattern recognition module is used to train and test partial discharge pattern recognition using an artificial neural network classifier that adopts a target adaptive genetic algorithm and utilizes fifteen-dimensional characteristic parameters.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the partial discharge pattern recognition method according to any one of claims 1 to 7 are implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the partial discharge pattern recognition method according to any one of claims 1 to 7 are implemented.

Images