SSA and RF-based GIS internal intermittent partial discharge pattern recognition system, method, electronic device and medium

Abstract

The application discloses a GIS internal intermittent discharge mode recognition system and method based on SSA and RF, acquires a PRPD two-dimensional graph of a GIS internal intermittent discharge signal and extracts characteristic parameters thereof; performs dimension reduction processing on the characteristic parameters, extracts principal components with contribution degrees higher than a set threshold value to form a characteristic vector group; finds optimal parameters by using a sparrow search algorithm to obtain a random forest optimized by the sparrow search algorithm; trains a random forest model by using a characteristic vector group of a training sample set, constructs an optimal GIS intermittent discharge fault diagnosis model, classifies a test sample set, and outputs a classification result of the test sample set. The application optimizes learning parameters in a random forest algorithm by using a sparrow search algorithm, solves the problem that parameters of a traditional random forest algorithm are difficult to select, and can more accurately and efficiently recognize GIS internal intermittent discharge fault types.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of power equipment fault type identification, and more particularly to a GIS internal intermittent discharge pattern recognition system, method, electronic device and medium based on SSA and RF. BACKGROUND

[0002] Gas insulated switchgear (GIS) is widely used in high-voltage and ultra-high-voltage fields due to its excellent insulation performance, small footprint, low maintenance workload, and strong environmental adaptability. Compared with conventional open-type substations, the maintenance interval of its main components is not less than 20 years, so it is often used in power plant booster stations to replace traditional substation switchgear and deliver power to transmission lines. Although GIS is generally considered to be low-maintenance or even maintenance-free, according to decades of operation experience, some insulation defects inevitably occur inside during manufacturing, transportation, installation, maintenance, and operation, such as metal burrs on conductors, loose or poor contact of components, air gaps formed by conductor and supporting insulators, leftovers after maintenance, and metal particles inside the cavity. These insulation defects will gradually deteriorate during long-term operation and eventually lead to insulation failure and power outage accidents. On the one hand, partial discharge will accelerate further damage to the internal insulation of the equipment, eventually leading to insulation failure and power outage accidents. For the phenomenon of intermittent discharge inside the ultra-high voltage GIS, the company organized technical experts to disassemble and analyze the fault, and the project leading unit, State Grid Electric Power Research Institute, participated in the fault analysis of some ultra-high voltage stations. It is believed that intermittent discharge is the cause of false positives and false negatives. In order to effectively ensure the safe and stable operation of GIS, GIS intermittent discharge fault diagnosis technology needs to be developed to ensure effective diagnosis of GIS intermittent discharge faults and safe and reliable operation of equipment.

[0003] With the development of artificial intelligence technology, existing technologies use machine learning algorithms to identify GIS partial discharge patterns. The most representative ones are artificial neural networks and support vector machines. Integrated algorithms such as random forest models based on bagging have high accuracy in predicting oil-dissolved gas. Machine learning algorithms have strong non-linear fitting capabilities and are widely used in partial discharge pattern recognition. However, the learning parameters in existing pattern recognition methods need to be determined manually, which requires a lot of time to select learning parameters. If the selection is not appropriate, it may lead to overfitting or local optimal solution problems. SUMMARY

[0004] To address the shortcomings of the prior art, the present application proposes a GIS internal intermittent discharge pattern recognition system and method based on SSA and RF to identify GIS internal intermittent discharge fault types.

[0005] To achieve this object, the GIS internal intermittent discharge mode recognition system based on SSA and RF designed by the application comprises a graph acquisition module, a feature parameter extraction module, a feature vector construction module, a random forest model construction module, a random forest model training module and a fault diagnosis module; the graph acquisition module is used to acquire two-dimensional graphs of GIS internal intermittent discharge signals, generate a fault sample label set, and divide the fault sample label set into a training sample set and a test sample set; the feature parameter extraction module is used to extract feature parameters of each two-dimensional graph in the fault sample label set, input the extracted feature parameters into the fault diagnosis module, and the fault diagnosis module determines the GIS internal intermittent discharge defect type; the feature vector construction module performs dimension reduction processing on the feature parameters output by the feature parameter extraction module, and selects feature parameters with contribution degree above a set threshold value from the dimension-reduced feature parameters as a feature vector group; the random forest model training module trains a random forest model (SSA and RF) optimized by the sparrow algorithm using the feature vector group of the training sample set, and constructs an optimal GIS intermittent discharge fault diagnosis model; the fault diagnosis module classifies the test sample set using the optimal GIS intermittent discharge fault diagnosis model constructed by the random forest model training module, and outputs the classification result of the test sample set.

[0006] A GIS internal intermittent discharge mode recognition method based on SSA and RF, which comprises the following steps:

[0007] Step 1: acquiring PRPD two-dimensional graphs of GIS internal intermittent discharge signals, generating a fault sample label set, and dividing the fault sample label set into a training sample set and a test sample set;

[0008] Step 2: extracting feature parameters of each two-dimensional graph in the fault sample label set, inputting the extracted feature parameters into a fault diagnosis module, and determining the GIS internal intermittent discharge defect type by the fault diagnosis module;

[0009] Step 3: performing dimension reduction processing on the feature parameters, and selecting feature parameters with contribution degree above a set threshold value from the dimension-reduced feature parameters as a feature vector group;

[0010] Step 4: training a random forest model optimized by the sparrow algorithm using the feature vector group of the training sample set, and constructing an optimal GIS intermittent discharge fault diagnosis model;

[0011] Step 5: classifying the test sample set using the optimal GIS intermittent discharge fault diagnosis model, and outputting the classification result of the test sample set.

[0012] The beneficial effects of this invention are as follows: This invention optimizes the learning parameters of random forest using the sparrow search algorithm, establishes SSA and RF models, solves the problem of difficult determination of learning parameters in traditional random forest algorithms, and improves the accuracy and efficiency of random forest. Attached Figure Description

[0013] Figure 1 This is a system structure diagram of the present invention;

[0014] Figure 2 This is a schematic diagram of the method flow of the present invention;

[0015] Figure 3 Diagram of the SSA optimization process;

[0016] Among them, 1-Graph acquisition module, 2-Feature parameter extraction module, 3-Feature vector construction module, 4-Random forest model construction module, 5-Random forest model training module, and 6-Fault diagnosis module. Detailed Implementation

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0018] A system for identifying intermittent discharge patterns inside a GIS based on SSA and RF, such as Figure 1 As shown, it includes a graph acquisition module 1, a feature parameter extraction module 2, a feature vector construction module 3, a random forest model construction module 4, a random forest model training module 5, and a fault diagnosis module 6;

[0019] The map acquisition module 1 is used to acquire the PRPD two-dimensional map of the intermittent discharge signal inside the GIS, generate a fault sample label set, and divide the fault sample label set into a training sample set and a test sample set.

[0020] The feature parameter extraction module 2 is used to extract the feature parameters of each two-dimensional map in the fault sample label set and perform normalization processing. The extracted feature parameters are input into the fault diagnosis module 6, and the fault diagnosis module 6 determines the type of intermittent discharge defect inside the GIS.

[0021] The feature vector construction module 3 uses principal component analysis to reduce the dimensionality of the feature parameters output by the feature parameter extraction module 2, and selects feature parameters with a contribution value above a set threshold from each feature parameter after dimensionality reduction as the feature vector group; in this embodiment, the preferred contribution value set threshold is 85%.

[0022] The random forest model construction module 4 is used to initialize the sparrow algorithm parameters, construct an initial random forest model, use the sparrow algorithm to find the optimal parameters, and obtain a random forest optimized by the sparrow algorithm.

[0023] The random forest model training module 5 trains the random forest model optimized by the sparrow algorithm using the feature vector group of the training sample set to construct an optimal GIS intermittent discharge fault diagnosis model.

[0024] The fault diagnosis module 6 classifies the test sample set by using the optimal GIS intermittent discharge fault diagnosis model constructed by the random forest model training module 5, determines the optimal classification result according to the voting result of all decision trees, and outputs the classification result of the test sample set.

[0025] In the technical solution, the two-dimensional atlas includes maximum discharge amount phase distribution Atlas, average discharge amount phase distribution Atlas, discharge frequency phase distribution Atlas and discharge amplitude distribution H n (q) atlas;

[0026] The intermittent discharge defect types include metal contamination, suspended electrode, insulator air gap defect, and metal particle defect.

[0027] In the technical solution, the characteristic parameters of the two-dimensional atlas include skewness, steepness, peak number, cross-correlation coefficient, discharge amount factor, and phase asymmetry.

[0028] In the technical solution, the specific implementation method of the random forest model construction module 4 is as follows:

[0029] Step 4.1, establish an initial random forest model, set the decision tree number and the maximum feature number value range;

[0030] Step 4.2, initialize the sparrow population parameters, set the population size N p , the maximum iteration number T max , the cross-validation fold V, the discoverer proportion PD, the proportion of sparrow aware of danger SD, and the safety threshold ST;

[0031] Step 4.3, select the mean square error MSE as the fitness function, calculate the fitness f of each sparrow, and the greater the fitness f, the better the position, and the global optimal position is the position with the maximum fitness value;

[0032] Step 4.4, update the discoverer position, and the discoverer position update formula of each generation is:

[0033]

[0034] In the formula, t is the current iteration number, i is the ith sparrow, iter max is the maximum iteration number; is the position of the ith sparrow in d dimensions in the t+1 iteration; is the position of the ith sparrow in d dimensions in the t iteration; a is a random number between (0, 1]; Q is a random number of normal distribution; R2 is an alarm value; ST is a security value; L is a matrix with size 1 x d and average value of 1;

[0035] The position of the joiner is updated, and the joiner position update formula is:

[0036]

[0037] In the formula, X p represents the current optimal position; x worst represents the current worst position; Q is a random number of normal distribution; i is the ith sparrow; n is the number of sparrows; x i,d represents the position of the ith sparrow in d dimensions; is the position of the ith sparrow in d dimensions in the t+1 iteration; is the position of the ith sparrow in d dimensions in the t iteration; is the best position of the t+1 iteration finder; A is a matrix of 1 x d, and each element is assigned a value of 1 or -1, and A + = A T (AA T ) -1 ;

[0038] The position of the alarm is updated, and the alarm position update formula is:

[0039]

[0040] In the formula, is the position of the ith sparrow in d dimensions in the t+1 iteration; is the position of the ith sparrow in d dimensions in the t iteration; β is a random number conforming to the standard normal distribution; x best represents the current global optimal position; represents the global best position of the t iteration; represents the global worst position of the t iteration; f i represents the current sparrow fitness value; f g represents the current best fitness value; f w represents the current worst fitness value; K represents a uniform random number in [-1, 1]; ε represents a small number to prevent the denominator from being 0;

[0041] 4.5, according to step 4.4, the positions of the finder, joiner and alarm are calculated, and if the new position fitness value of the finder, joiner or alarm is better than the previous finder, joiner or alarm position, the global optimal position is updated;

[0042] 4.6, check whether the maximum number of iterations is met, if the maximum number of iterations is not met, jump to 4.4 to continue execution, if the maximum number of iterations is met, output the optimal parameters.

[0043] The embodiment also provides a GIS internal intermittent discharge mode recognition method based on SSA and RF, as shown in the figure, which comprises the following steps: Figure 2

[0044] Step 1, acquire PRPD two-dimensional maps of GIS internal intermittent discharge signals, generate a fault sample label set, and divide the fault sample label set into a training sample set and a test sample set;

[0045] Step 2, extract feature parameters of each two-dimensional map in the fault sample label set, input the extracted feature parameters into a fault diagnosis module 6, and the fault diagnosis module 6 determines a GIS internal intermittent discharge defect type;

[0046] Step 3, perform dimension reduction processing on the feature parameters, and select feature parameters with contribution degrees higher than a set threshold value as a feature vector group from the feature parameters after the dimension reduction processing;

[0047] Step 4, initialize sparrow algorithm parameters, construct an initial random forest model, find optimal parameters by using the sparrow algorithm, and obtain a sparrow algorithm optimized random forest;

[0048] Step 5, train the sparrow algorithm optimized random forest model by using the feature vector group of the training sample set, and construct an optimal GIS intermittent discharge fault diagnosis model;

[0049] Step 6, perform classification processing on the test sample set by using the optimal GIS intermittent discharge fault diagnosis model, and output a classification result of the test sample set.

[0050] The embodiment also provides an electronic device comprising a processor, and the processor is used to execute program instructions to realize the GIS internal intermittent discharge mode recognition method based on SSA and RF provided in the above embodiment.

[0051] The embodiment also provides a storage medium comprising a processor, and the storage medium stores program instructions used to realize the GIS internal intermittent discharge mode recognition method based on SSA and RF provided in the above embodiment.

[0052] ​Those aspects of the specification that are not otherwise fully described are deemed to be part of the prior art. Those skilled in the art will appreciate that embodiments of the present application can be practiced in a method, system, or computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, magnetic disks, CD-ROMs, optical storage media such as DVD s, etc.) embodying computer readable program code.

[0053] The present application is described in reference to the flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flow diagrams and / or block diagrams, and combinations of blocks in the flow diagrams and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing device or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flow diagrams and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flow diagram and / or block diagram block or blocks. Figure 1 means for performing each of the functions specified in the flow diagram and / or block diagram block or blocks.

[0054] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flow diagrams and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flow diagram and / or block diagram block or blocks. Figure 1 means for performing each of the functions specified in the flow diagram and / or block diagram block or blocks.

[0055] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow diagrams and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flow diagram and / or block diagram block or blocks. Figure 1 means for performing each of the functions specified in the flow diagram and / or block diagram block or blocks.

[0056] It should be pointed out finally that the above examples are only used for illustrating the technical solutions of the present application but not for limiting the protection scope thereof, and although the present application has been described in detail with reference to the above examples, it should be understood by those skilled in the art that the specific embodiments of the present application can be changed, modified or replaced equivalently by those skilled in the art after reading the present application, but these changes, modifications or equivalent replacements are all within the protection scope of the claims to be filed of the present application.

Claims

1. A system for identifying intermittent discharge patterns inside a GIS based on SSA and RF, characterized in that, include: The system comprises a graph acquisition module (1), a feature parameter extraction module (2), a feature vector construction module (3), a random forest model construction module (4), a random forest model training module (5), and a fault diagnosis module (6). The map acquisition module (1) is used to acquire a two-dimensional map of the intermittent discharge signal inside the GIS, generate a fault sample label set, and divide the fault sample label set into a training sample set and a test sample set. The feature parameter extraction module (2) is used to extract the feature parameters of each two-dimensional map in the fault sample label set, and input the extracted feature parameters into the fault diagnosis module (6) so that the fault diagnosis module (6) can determine the type of intermittent discharge defect inside the GIS based on the feature parameters. The feature vector construction module (3) performs dimensionality reduction processing on the feature parameters output by the feature parameter extraction module (2), and selects feature parameters with a contribution setting threshold above each feature parameter after dimensionality reduction as a feature vector group; Random forest model construction module (4), the random forest model construction module (4) is used to initialize the sparrow algorithm parameters, construct an initial random forest model, use the sparrow algorithm to find the optimal parameters, and obtain a random forest optimized by the sparrow algorithm; The random forest model training module (5) uses the feature vector group of the training sample set to train the random forest model optimized by the sparrow algorithm, and constructs the optimal GIS intermittent discharge fault diagnosis model. The fault diagnosis module (6) uses the optimal GIS intermittent discharge fault diagnosis model constructed by the random forest model training module (5) to classify the test sample set and output the classification result of the test sample set.

2. The GIS internal intermittent discharge pattern recognition system based on SSA and RF as described in claim 1, characterized in that: The two-dimensional spectrum includes the phase distribution of maximum discharge. Spectrum, Phase distribution of average discharge quantity Spectrum, phase distribution of discharge times Spectrum and discharge amplitude distribution H n (q) graph.

3. The GIS internal intermittent discharge pattern recognition system based on SSA and RF as described in claim 1, characterized in that: The intermittent discharge defect types include metal contamination, suspended electrodes, insulator air gap defects, and metal particle defects.

4. The GIS internal intermittent discharge pattern recognition system based on SSA and RF as described in claim 1, characterized in that: The characteristic parameters of the two-dimensional spectrum include skewness, steepness, number of peaks, cross-correlation coefficient, discharge factor, and phase asymmetry.

5. The GIS internal intermittent discharge pattern recognition system based on SSA and RF as described in claim 1, characterized in that: The feature vector construction module (3) uses principal component analysis for dimensionality reduction.

6. The GIS internal intermittent discharge pattern recognition system based on SSA and RF as described in claim 1, characterized in that, The specific implementation method of the random forest model construction module (4) is as follows: Step 4.1: Establish an initial random forest model, and set the range of values ​​for the number of decision trees and the maximum number of features in the initial random forest model; Step 4.2: Initialize the sparrow population parameters in the initial random forest model. The initialization of sparrow population parameters includes the population size N. p Maximum number of iterations T max Cross-validation folds V, discoverer ratio PD, percentage of sparrows aware of danger SD, and safety threshold ST; Step 4.3: Select the mean squared error (MSE) as the fitness function and calculate the fitness f for each sparrow; Step 4.4, update the discoverer's location. The formula for updating the discoverer's location for each generation is: In the formula, t is the current iteration number; i is the i-th sparrow; iter max It is the maximum number of iterations; It is the position of the i-th sparrow in the t+1 iteration of the d-dimensional array; is the position of the i-th sparrow in t iterations in d dimensions; α is a random number between (0,1]; Q is a normally distributed random number; R² is the warning value; ST is the safety value; L is a matrix of size 1×d with an element mean of 1; The formula for updating joiner positions is as follows: In the formula, x worst This represents the worst-case position; Q is a normally distributed random number; i is the i-th sparrow; n is the number of sparrows; x i,d This represents the position of the i-th sparrow in dimension d; It is the position of the i-th sparrow in the t+1 iteration of the d-dimensional array; The position of the i-th sparrow in d-dimensional t iterations; It is the optimal position of the discoverer in iteration t+1; A is a 1×d matrix where each element is assigned the value 1 or -1, and A + =A T (AA T ) -1 ; Update the vigilant's position using the following formula: In the formula, The position of the i-th sparrow in the t+1 iteration of the d-dimensional array; The position of the i-th sparrow in d-dimensional iteration t; β is a random number that follows a standard normal distribution; x best This indicates the current globally optimal position; This represents the globally optimal position in the t-th iteration; f represents the worst-case position in the t-th iteration; i This represents the current sparrow fitness value; f g f represents the current best fitness value; w This represents the current worst fitness value; K represents a uniformly random number in the range [-1, 1]; ε represents a small number to prevent the denominator from being 0. 4.5 Calculate the positions of the discoverer, joiner, and vigilant according to step 4.

4. If the fitness value of the new position of the discoverer, joiner, or vigilant is better than the previous position of the discoverer, joiner, or vigilant, then update the global optimal position. 4.6 Check if the maximum number of iterations is met. If the maximum number of iterations is not met, jump to 4.4 and continue execution. If the maximum number of iterations is met, output the optimal parameters. The optimal parameters will be used to adjust the parameters in the random forest model to obtain the random forest model optimized by the Sparrow Algorithm.

7. A method for identifying intermittent discharge patterns inside a GIS based on SSA and RF using the system described in claim 1, characterized in that: It includes the following steps: Step 1: Obtain the PRPD two-dimensional map of the intermittent discharge signal inside the GIS, generate a fault sample label set, and divide the fault sample label set into a training sample set and a test sample set; Step 2: Extract the feature parameters of each two-dimensional map in the fault sample label set, and input the extracted feature parameters into the fault diagnosis module (6). The fault diagnosis module (6) determines the type of intermittent discharge defect inside the GIS. Step 3: The feature parameters are subjected to dimensionality reduction processing, and feature parameters with a contribution value above a set threshold are selected from the dimensionality-reduced feature parameters as feature vector groups. Step 4: Initialize the sparrow algorithm parameters, construct the initial random forest model, use the sparrow algorithm to find the optimal parameters, and obtain the random forest optimized by the sparrow algorithm; Step 5: Use the feature vector group of the training sample set to train the random forest model optimized by the sparrow algorithm to construct the optimal GIS intermittent discharge fault diagnosis model; Step 6: Use the optimal GIS intermittent discharge fault diagnosis model to classify the test sample set and output the classification results of the test sample set.

8. An electronic device, characterized in that, Includes a processor for executing program instructions to implement the SSA and RF-based method for identifying intermittent discharge patterns within a GIS as described in claim 7.

9. A storage medium, characterized in that, The storage medium stores a method for implementing the intermittent discharge pattern recognition method for GIS based on SSA and RF as described in claim 7.

Images