Fault diagnosis method, device and equipment for gas insulated switchgear and medium

Abstract

The invention discloses a fault diagnosis method, device, equipment and medium for a gas insulated switchgear, and relates to the technical field of power systems, and the method comprises the steps: determining a sample set based on a sulfur hexafluoride decomposition component data set, carrying out the training of a base classifier based on the sample set and a Bagging algorithm, and carrying out the fault diagnosis of a gas insulated switchgear. Determining each preliminary diagnosis result based on the sample test set and the obtained trained base classifier; converting the preliminary diagnosis result into a basic probability assignment, judging whether a current scene is a scene meeting a preset high-conflict condition, improving a Dempster synthesis rule by using a Dubois-Prade rule based on a judgment result, and determining a target fault diagnosis model by using the improved Dempster synthesis rule; and determining a target diagnosis result based on a to-be-diagnosed data set corresponding to the to-be-detected gas insulated switchgear by using the target fault diagnosis model. And faults in the GIS can be quickly and effectively diagnosed, and the fault types can be identified.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a fault diagnosis method, apparatus, equipment and medium for gas-insulated combined electrical appliances. Background Technology

[0002] Sulfur hexafluoride (SF6) gas is widely used in ultra-high voltage insulation media due to its excellent insulating and arc-quenching properties. However, when SF6 gas occurs in insulating equipment during partial discharge (PD) faults, spark discharges, and partial overthermal faults (POF), it may decompose to form stable products such as CO2, SO2, SOF2, CF4, H2S, SO2F2, and SOF4. The content and proportion of these decomposition components are closely related to the type of insulation fault.

[0003] Currently, research on fault diagnosis methods for sulfur hexafluoride decomposition components in GIS (Gas Insulated Switchgear) mainly relies on single machine learning algorithms, which are often limited in practical applications due to low accuracy or limited universality. In particular, regardless of the classification algorithm used, the identification accuracy for P-type defects (i.e., free metal particles) and M-type defects (i.e., insulator contamination) is low. This is because the data content for P-type and M-type defects is highly dispersed, and the data characteristics are not clearly defined, easily leading to errors in machine learning judgments. Furthermore, when applied to high-dimensional data, obtaining the joint distribution of features after quickly obtaining the conditional probability distribution of features becomes extremely difficult, and there are certain errors in manually calculating relevant parameters (e.g., calculating termination conditions, selecting step sizes, etc.).

[0004] As can be seen from the above, how to quickly and effectively diagnose faults within GIS and identify their types is an urgent problem to be solved. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a fault diagnosis method, apparatus, equipment, and medium for gas-insulated switchgear (GIS), which can quickly and effectively diagnose faults within GIS and identify their fault types. The specific solution is as follows:

[0006] In a first aspect, this application provides a fault diagnosis method for gas-insulated switchgear, including:

[0007] The target number of base classifiers is determined using a pre-set test experiment and a sulfur hexafluoride decomposition component dataset. Based on the sulfur hexafluoride decomposition component dataset and using sampling with replacement, a sample set corresponding to the target number is determined. The base classifiers for the target number are trained using the sample set and the Bagging algorithm to obtain trained base classifiers. The sample test set is input into the trained base classifiers to obtain preliminary diagnostic results. The sulfur hexafluoride decomposition component dataset includes sulfur hexafluoride decomposition components and fault types.

[0008] The preliminary diagnosis results are converted into basic probability values, and the current scene is judged based on the basic probability values ​​to determine whether the scene meets the preset high conflict conditions, so as to obtain the corresponding judgment results.

[0009] Based on the judgment result, the Dempster synthesis rule is improved using the Dubois-Prade rule to obtain the improved Dempster synthesis rule. The basic probability assignment is fused using the improved Dempster synthesis rule, and the target fault diagnosis model is determined based on the fusion result.

[0010] Obtain the diagnostic dataset corresponding to the internal structure of the gas-insulated switchgear under test, and determine the target diagnostic result, including fault type and fault confidence, based on the diagnostic dataset and the target fault diagnosis model; the diagnostic dataset is the sulfur hexafluoride decomposition component corresponding to the internal structure of the gas-insulated switchgear under test.

[0011] Optionally, determining the target number of the base classifier using a preset test experiment and a dataset of sulfur hexafluoride decomposition components includes:

[0012] The dataset of sulfur hexafluoride decomposition components is divided according to a preset division ratio to obtain a training set and a test set.

[0013] Based on the training set, several sample sets are constructed using sampling with replacement. A corresponding number of base classifiers are trained using the sample sets. The voting rules of the Bagging algorithm are used to combine the output results of the base classifiers to obtain the ensemble result.

[0014] Based on the integration results, the accuracy of the training set and the accuracy of the test set are compared with the first real fault label in the training set and the second real fault label in the test set, respectively, so as to determine the accuracy of the training set and the accuracy of the test set using the comparison results.

[0015] The number of base classifiers that meet the preset comprehensive performance conditions is determined by using the training set accuracy and test set accuracy corresponding to different numbers of base classifiers, and the number of base classifiers is determined as the target number.

[0016] Optionally, the step involves determining a sample set corresponding to the target quantity based on the sulfur hexafluoride decomposition component dataset and using sampling with replacement; training a base classifier for the target quantity based on the sample set and the Bagging algorithm to obtain a trained base classifier; and inputting the sample test set into the trained base classifier to obtain preliminary diagnostic results, including:

[0017] Based on the sulfur hexafluoride decomposition component dataset, a sample set is obtained by repeatedly sampling with replacement a number of times corresponding to the target quantity; the sample set has the same length as the sulfur hexafluoride decomposition component dataset.

[0018] The base classifier is trained using the Bagging algorithm and the sample set to obtain each trained base classifier. The sample test set is then input into the trained base classifier to obtain each preliminary diagnostic result. The sample test set is a dataset determined based on the sulfur hexafluoride decomposition component dataset.

[0019] Optionally, the step of converting the preliminary diagnosis result into a basic probability assignment, and determining whether the current scenario meets the preset high-conflict conditions based on the basic probability assignment, to obtain the corresponding judgment result, includes:

[0020] The preliminary diagnostic results are converted into basic probability assignments based on preset constraints; the preset constraints include that the confidence level of the empty set in the basic probability assignment is a first preset value, and that the total confidence level in the basic probability assignment is a second preset value.

[0021] Determine the conflict coefficient between each of the basic probability assignments, and determine whether the conflict coefficient is not less than a first preset threshold.

[0022] If the conflict coefficient is not less than the first preset threshold, then the current scenario is characterized as a scenario that meets the preset high conflict condition.

[0023] Optionally, after converting the preliminary diagnostic results into basic probability values ​​based on preset constraints, the method further includes:

[0024] Determine the criterion value between each of the basic probability assignments, and determine whether the criterion value is not less than a second preset threshold.

[0025] If the criterion value is not less than the second preset threshold, then the current scenario is characterized as a scenario that meets the preset high conflict condition.

[0026] Optionally, the step of improving the Dempster synthesis rule based on the judgment result and using the Dubois-Prade rule to obtain an improved Dempster synthesis rule, fusing the basic probability assignments using the improved Dempster synthesis rule, and determining the target fault diagnosis model based on the fusion result includes:

[0027] If the current scenario meets the preset high conflict conditions, the Dempster synthesis rule is improved by using the Dubois-Prade rule to obtain the improved Dempster synthesis rule. The basic probability assignment is fused using the improved Dempster synthesis rule, and the target fault diagnosis model is determined based on the fusion result.

[0028] If the current scenario does not meet the preset high-conflict condition, the basic probability assignments are fused using the Dempster synthesis rule, and the output result of the target fault diagnosis model is determined based on the fusion result.

[0029] Optionally, the step of acquiring the diagnostic dataset corresponding to the internal structure of the gas-insulated switchgear under test, and determining the target diagnostic result including fault type and fault confidence based on the diagnostic dataset and the target fault diagnosis model, includes:

[0030] A diagnostic dataset is constructed based on the decomposition components of sulfur hexafluoride, the insulating medium, when a fault occurs inside the gas-insulated switchgear under test.

[0031] The target fusion result corresponding to the dataset to be diagnosed is determined using the target fault diagnosis model. The target diagnosis result, including fault type and fault confidence, is determined based on the basic probability assignment with the largest value in the target fusion result.

[0032] Secondly, this application provides a fault diagnosis device for gas-insulated switchgear, comprising:

[0033] The base classifier training module is used to determine the target number of base classifiers using a preset test experiment and a sulfur hexafluoride decomposition component dataset. Based on the sulfur hexafluoride decomposition component dataset and using sampling with replacement, a sample set corresponding to the target number is determined. Based on the sample set and the Bagging algorithm, the base classifiers for the target number are trained to obtain trained base classifiers. The sample test set is input into the trained base classifiers to obtain preliminary diagnostic results. The sulfur hexafluoride decomposition component dataset includes sulfur hexafluoride decomposition components and fault types.

[0034] The scene judgment module is used to convert the preliminary diagnosis result into a basic probability assignment, and to judge whether the current scene meets the preset high conflict conditions based on the basic probability assignment, so as to obtain the corresponding judgment result.

[0035] The diagnostic model determination module is used to improve the Dempster synthesis rule based on the judgment result and using the Dubois-Prade rule to obtain the improved Dempster synthesis rule, and to fuse the basic probability assignment using the improved Dempster synthesis rule, and to determine the target fault diagnosis model based on the fusion result.

[0036] The diagnostic result determination module is used to acquire the diagnostic dataset corresponding to the internal components of the gas-insulated switchgear under test, and to determine the target diagnostic result, including the fault type and fault confidence, based on the diagnostic dataset and the target fault diagnosis model; the diagnostic dataset is the sulfur hexafluoride decomposition component corresponding to the internal components of the gas-insulated switchgear under test.

[0037] Thirdly, this application provides an electronic device, comprising:

[0038] Memory, used to store computer programs;

[0039] A processor is used to execute the computer program to implement the aforementioned fault diagnosis method for gas-insulated switchgear.

[0040] Fourthly, this application provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned fault diagnosis method for gas-insulated combined electrical appliances.

[0041] This application utilizes a pre-set test experiment and a sulfur hexafluoride (SF6) decomposition component dataset to determine the target number of base classifiers. Based on the SF6 decomposition component dataset and using sampling with replacement, a sample set corresponding to the target number is determined. The base classifiers for the target number are trained using the sample set and the Bagging algorithm to obtain trained base classifiers. The sample test set is input into the trained base classifiers to obtain preliminary diagnostic results. The SF6 decomposition component dataset includes SF6 decomposition components and fault types. The preliminary diagnostic results are converted into basic probability assignments, and based on these basic probability assignments, it is determined whether the current scenario meets the pre-set high-conflict conditions. The scene is used to obtain the corresponding judgment result; based on the judgment result, the Dempster synthesis rule is improved using the Dubois-Prade rule to obtain the improved Dempster synthesis rule; the basic probability assignment is fused using the improved Dempster synthesis rule; the target fault diagnosis model is determined based on the fusion result; the diagnostic dataset corresponding to the inside of the gas-insulated switchgear under test is obtained; based on the diagnostic dataset and the target fault diagnosis model, the target diagnosis result including fault type and fault confidence is determined; the diagnostic dataset is the sulfur hexafluoride decomposition component corresponding to the inside of the gas-insulated switchgear under test.

[0042] As can be seen from the above, this application tests the performance of different numbers of base classifiers through pre-set test experiments and sulfur hexafluoride decomposition component datasets to determine the target number. This avoids computational redundancy caused by too many base classifiers and insufficient diversity caused by too few base classifiers. Based on the sulfur hexafluoride decomposition component dataset, sample sets are generated by sampling with replacement. Each sample set differs due to the randomness of sampling. Then, the corresponding base classifier is trained using the Bagging algorithm. The preliminary diagnostic results corresponding to the base classifiers are converted into basic probability assignments. The diagnostic confidence of the preliminary diagnostic results of the base classifiers is quantified to avoid misjudgment. High-conflict scenarios are identified based on the basic probability assignments, i.e., the preliminary diagnostic results of base classifier 1 are completely opposite to those of base classifier 2. The Dempster synthesis rule is improved by using the Dubois-Prade rule, which significantly improves the diagnostic accuracy of the target fault diagnosis model in high-conflict scenarios. In this way, by inputting the diagnostic dataset corresponding to the gas-insulated switchgear under test into the target fault diagnosis model, it is possible not only to accurately locate the fault type inside the gas-insulated switchgear under test, but also to ensure the reliability of the diagnostic results, improve the efficiency and generalization ability of insulation defect fault identification, and thus quickly investigate and deal with the fault, and maintain the safe and stable operation of the power system. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0044] Figure 1 This is a flowchart of a fault diagnosis method for a gas-insulated combined electrical appliance disclosed in this application;

[0045] Figure 2 A schematic diagram illustrating the comparison of training and test set accuracy under different numbers of base classifiers provided in this application; Figure 2 (a) is a schematic diagram comparing the accuracy of training and test sets for N types of insulation defects with different numbers of base classifiers; Figure 2 (b) is a schematic diagram comparing the accuracy of training and test sets for P-type insulation defects with different numbers of base classifiers; Figure 2 (c) is a schematic diagram comparing the accuracy of training and test sets for M-type insulation defects with different numbers of base classifiers; Figure 2 (d) is a schematic diagram comparing the accuracy of training and test sets for G-type insulation defects with different numbers of base classifiers;

[0046] Figure 3 A flowchart of a traditional Bagging algorithm is provided in this application;

[0047] Figure 4 A flowchart of an improved Bagging algorithm provided for this application;

[0048] Figure 5 A comparative diagram illustrating the impact of different synthesis methods on the accuracy of the model test set, provided in this application;

[0049] Figure 6 A comparative diagram illustrating the impact of different synthesis methods on the precision, recall, and F1 score of the model test set, as provided in this application;

[0050] Figure 7 A schematic diagram illustrating the diagnostic performance of a target fault diagnosis model provided in this application;

[0051] Figure 8 A schematic diagram showing the comparison of the diagnostic performance of the Bagging algorithm before and after the improvement provided in this application; Figure 8 (a) is a schematic diagram comparing the diagnostic performance of the Bagging algorithm before its improvement; Figure 8 (b) is a schematic diagram comparing the diagnostic performance of an improved Bagging algorithm;

[0052] Figure 9 This is a schematic diagram of the fault diagnosis device for a gas-insulated combined electrical appliance disclosed in this application;

[0053] Figure 10 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] Currently, research on fault diagnosis methods for sulfur hexafluoride decomposition components within GIS (Gas Insulated Switchgear) mainly relies on single machine learning algorithms, which are often limited in practical applications due to low accuracy or limited universality. In particular, regardless of the classification algorithm used, the identification accuracy for P-type and M-type defects is low. Therefore, this application provides a fault diagnosis method for gas-insulated switchgear. The method inputs the diagnostic dataset corresponding to the internal structure of the gas-insulated switchgear under test into the target fault diagnosis model. This not only accurately locates the fault type within the gas-insulated switchgear but also ensures the reliability of the diagnostic results, improving the efficiency and generalization ability of insulation defect identification. This allows for rapid fault investigation and handling, maintaining the safe and stable operation of the power system.

[0056] See Figure 1 As shown in the figure, an embodiment of the present invention discloses a fault diagnosis method for gas-insulated switchgear, comprising:

[0057] Step S11: Determine the target number of base classifiers using a preset test experiment and a sulfur hexafluoride decomposition component dataset. Based on the sulfur hexafluoride decomposition component dataset and using sampling with replacement, determine the sample set corresponding to the target number. Train the base classifiers for the target number based on the sample set and the Bagging algorithm to obtain trained base classifiers. Input the sample test set into the trained base classifiers to obtain preliminary diagnostic results. The sulfur hexafluoride decomposition component dataset includes sulfur hexafluoride decomposition components and fault types.

[0058] In this embodiment, based on A sulfur hexafluoride (SF6) decomposition component dataset is constructed by decomposing the products and corresponding fault types. This dataset is then divided into training and testing sets. Based on the training set, several sample sets are constructed using sampling with replacement (sampling with replacement followed by replacement). The performance of each base classifier is then tested using the sample sets. The training and testing accuracy rates for four types of insulation defects are statistically analyzed using a Bagging (Bootstrap aggregating) algorithm with voting. These four types of insulation defects include: N-type (metal protrusions); P-type (free metal particles); M-type (insulator contamination); and G-type (insulator air gaps). In one specific embodiment… Figure 2 This embodiment provides a schematic diagram comparing the accuracy of the training and test sets under different numbers of base classifiers. Figure 2 (a) is a schematic diagram comparing the accuracy of training and test sets for N types of insulation defects with different numbers of base classifiers; Figure 2 (b) is a schematic diagram comparing the accuracy of training and test sets for P-type insulation defects with different numbers of base classifiers; Figure 2 (c) is a schematic diagram comparing the accuracy of training and test sets for M-type insulation defects with different numbers of base classifiers; Figure 2 (d) is a schematic diagram comparing the accuracy of training and test sets for class G insulation defects with different numbers of base classifiers; from Figure 2 A comparison shows that when the number of base classifiers is 12-16, the accuracy on the training and test sets for various defects is the most stable and highest. Therefore, the target number of base classifiers can be 12-16, preferably 14. It is worth mentioning that the base classifiers in this embodiment can be base classifiers determined based on MCMC (Markov Chain Monte Carlo Method) combined with GPC (Generalized Predictive Control).

[0059] Specifically, determining the target number of base classifiers using a preset test experiment and a sulfur hexafluoride decomposition component dataset includes: dividing the sulfur hexafluoride decomposition component dataset according to a preset partitioning ratio to obtain a training set and a test set; constructing several sample sets based on the training set using sampling with replacement, training a corresponding number of base classifiers using the sample sets, and combining the output results of the base classifiers using the voting rule of the Bagging algorithm to obtain an ensemble result; comparing the ensemble result with the first true fault label in the training set and the second true fault label in the test set to determine the training set accuracy and the test set accuracy using the comparison results; determining the number of base classifiers that meet the preset comprehensive performance conditions using the training set accuracy and test set accuracy corresponding to different numbers of base classifiers, and determining the number of base classifiers as the target number.

[0060] It is understood that after obtaining the target quantity, the sample set is obtained by repeatedly sampling the number of times corresponding to the target quantity based on the sulfur hexafluoride decomposition component dataset and using sampling with replacement. Then, the base classifier is trained using the sample set based on the Bagging algorithm to obtain each trained base classifier. The sample test set is then input into the trained base classifier to obtain each preliminary diagnostic result. Specifically, the process involves determining a sample set corresponding to the target quantity based on the sulfur hexafluoride decomposition component dataset using sampling with replacement, training a base classifier for the target quantity using the sample set and the Bagging algorithm to obtain a trained base classifier, and inputting a sample test set into the trained base classifier to obtain preliminary diagnostic results. This includes: repeatedly sampling the target quantity a number of times using sampling with replacement based on the sulfur hexafluoride decomposition component dataset to obtain a sample set; the sample set has the same length as the sulfur hexafluoride decomposition component dataset; training the base classifier using the Bagging algorithm and the sample set to obtain trained base classifiers, and inputting a sample test set into the trained base classifier to obtain preliminary diagnostic results; the sample test set is a dataset determined based on the sulfur hexafluoride decomposition component dataset.

[0061] Step S12: Convert the preliminary diagnosis result into a basic probability assignment, and determine whether the current scene meets the preset high conflict conditions based on the basic probability assignment, so as to obtain the corresponding judgment result.

[0062] In this embodiment, in the Bagging algorithm, after obtaining each preliminary diagnostic result, the final target diagnostic result is determined based on each preliminary diagnostic result and using a voting method. Figure 3This embodiment provides a flowchart of a traditional Bagging algorithm. The sample determination method is random sampling. After obtaining the preliminary diagnostic results, a voting method is used to determine the final diagnostic model, which in turn yields the target diagnostic result. The corresponding formula is as follows:

[0063] ;

[0064] in, This represents the target diagnostic result corresponding to the voting-based diagnostic model. The number of base classifiers; Fault category; As a base classifier; This is an indicator function.

[0065] It is understood that the target fault diagnosis model obtained in this embodiment does not use the voting method, but instead uses the decision fusion method. Figure 4 This embodiment provides an improved Bagging algorithm flowchart. The sample determination method is bootstrapping, i.e., sampling with replacement, and the voting method is replaced with a decision fusion method. Specifically, after obtaining the preliminary diagnosis result, the preliminary diagnosis result is converted into basic probability assignments in Dempster-Shafer evidence theory. Dempster-Shafer evidence theory defines... Define an idempotent set, which represents the entire set of all possible failure types. For all combinations of subsets of the entire set, i.e., all possible combinations of diagnostic conclusions, the basic probability assignment has two constraints. Here, an empty set represents the absence of faults, i.e., a confidence value of 0. This indicates that the sum of the confidence values ​​of all possible conclusions is 1, meaning that there will not be a situation where the classifier classifies 80% of cases as P-class defects and 30% as M-class defects. Then, based on the basic probability assignment, it is determined whether the current scene meets the preset high-conflict condition. If yes, the process jumps to the step of improving the Dempster synthesis rule using the Dubois-Prade rule to obtain the improved Dempster synthesis rule, and then fusing the basic probability assignment using the improved Dempster synthesis rule; otherwise, the basic probability assignment is directly fused using the Dempster synthesis rule. Further, a conflict coefficient is determined based on the basic probability assignment, and it is determined whether the conflict coefficient is not less than a first preset threshold. If the conflict coefficient is not less than the first preset threshold, it indicates that the current scene meets the preset high-conflict condition. The formula corresponding to the conflict coefficient is as follows:

[0066] ;

[0067] in, The conflict coefficient is mentioned above; The initial diagnostic results are as described above.

[0068] Specifically, the step of converting the preliminary diagnosis result into basic probability assignments and determining whether the current scenario meets preset high conflict conditions based on the basic probability assignments to obtain a corresponding judgment result includes: converting the preliminary diagnosis result into basic probability assignments based on preset constraints; the preset constraints include the empty set confidence level in the basic probability assignments being a first preset value, and the total confidence level in the basic probability assignments being a second preset value; determining the conflict coefficient between each of the basic probability assignments, and determining whether the conflict coefficient is not less than a first preset threshold; if the conflict coefficient is not less than the first preset threshold, then the current scenario is characterized as meeting preset high conflict conditions.

[0069] Furthermore, a criterion value can be determined based on the basic probability assignment, and the criterion value can be used to determine whether the current scenario meets the preset high-conflict condition. In one specific implementation, the criterion value is a precision fluctuation threshold, which is a stable criterion value. If the criterion value is not less than 0.0001, it indicates that the current scenario meets the preset high-conflict condition. Specifically, after converting the preliminary diagnosis result into a basic probability assignment based on the preset constraint condition, the method further includes: determining the criterion value between each of the basic probability assignments, and determining whether the criterion value is not less than a second preset threshold; if the criterion value is not less than the second preset threshold, it indicates that the current scenario meets the preset high-conflict condition. It is worth mentioning that the first preset threshold and the second preset threshold can be adjusted according to the actual situation, and are not specifically limited here.

[0070] Step S13: Based on the judgment result, the Dempster synthesis rule is improved using the Dubois-Prade rule to obtain the improved Dempster synthesis rule. The basic probability assignment is fused using the improved Dempster synthesis rule, and the target fault diagnosis model is determined based on the fusion result.

[0071] In this embodiment, if the current scenario meets the preset high conflict conditions, the Dempster composition rule is improved to address the limitations of conflict scenarios. Figure 5This diagram illustrates the comparison of the impact of different synthesis methods on the accuracy of the model on the test set, as provided in this embodiment. The horizontal axis represents the criterion value, ranging from 0 to 1, including minimum values ​​such as 1E-7. The vertical axis represents the accuracy. The four curves represent four synthesis methods: DS represents the traditional Dempster synthesis rule; DS-Yager represents the Yager improvement method in Dempster-Shafer; DS-Sun Quan represents the Sun Quan improvement method in Dempster-Shafer; and Dubois and Prade represents the Dubois-Prade improvement method in Dempster-Shafer. It can be seen that the Dubois-Prade improvement method has the highest accuracy. Furthermore, Figure 6 This is a comparative diagram illustrating the impact of different synthesis methods on the precision, recall, and F1 score of the model on the test set, provided in this embodiment. Figure 5 and Figure 6 It can be seen that the Dubois-Prade method has the most stable and high precision, recall and F1 score. Therefore, the Dubois-Prade rule (i.e. the Dubois-Prade combination rule) is selected to improve the Dempster synthesis rule to obtain the improved Dempster synthesis rule.

[0072] Understandably, the core logic of the Dubois-Prade rule is to first extract consensus information and then reasonably allocate conflicting information, as shown in the following formula:

[0073] ;

[0074] in, Assign values ​​to the basic probabilities; , Assign values ​​to the two basic probabilities to be merged; , They are respectively Corresponding subsets and The corresponding subset; The complete set defined for the DS evidence theory; The conflict coefficient; and For the reconcilable conflict portion, in which, ; .

[0075] The formulas corresponding to Dempster's synthesis rules are as follows:

[0076] ;

[0077] in, A subset of the basic probabilities assigned to each preliminary diagnostic result; The number of base classifiers; Let be the conflict coefficient.

[0078] In this embodiment, the improved Dempster synthesis rule is used to fuse the basic probability assignments, and the target fault diagnosis model is determined based on the fusion result. Figure 7 This diagram illustrates the diagnostic performance of a target fault diagnosis model provided in this embodiment. The horizontal axis represents four types of defects, and the vertical axis represents the diagnostic accuracy of the target fault diagnosis model for the four types of defects. It can be seen that the target fault diagnosis model has a high diagnostic accuracy for all four types of defects. Furthermore, if the current scenario does not meet the preset high-conflict condition, the basic probability assignments are directly fused using Dempster's synthesis rules, and the output result of the target fault diagnosis model is determined based on the fusion result.

[0079] Specifically, the step of improving the Dempster synthesis rule based on the judgment result and using the Dubois-Prade rule to obtain an improved Dempster synthesis rule, fusing the basic probability assignments using the improved Dempster synthesis rule, and determining the target fault diagnosis model based on the fusion result includes: if the current scenario meets the preset high conflict conditions, then improving the Dempster synthesis rule using the Dubois-Prade rule to obtain an improved Dempster synthesis rule, fusing the basic probability assignments using the improved Dempster synthesis rule, and determining the target fault diagnosis model based on the fusion result; if the current scenario does not meet the preset high conflict conditions, then fusing the basic probability assignments using the Dempster synthesis rule, and determining the output result of the target fault diagnosis model based on the fusion result.

[0080] Figure 8 This is a schematic diagram comparing the diagnostic performance of the Bagging algorithm before and after the improvement, as provided in this embodiment. Figure 8 (a) A schematic diagram comparing the diagnostic performance of the Bagging algorithm before the improvement. Figure 8 (b) is a diagram comparing the diagnostic performance of the improved Bagging algorithm. The horizontal axis represents various types of defects, and the vertical axis represents the corresponding F1 score, precision, and recall. It can be seen that replacing the voting method in the original Bagging algorithm with the decision fusion method can effectively improve the diagnostic precision, especially in the identification accuracy of P-type and M-type defects. The identification accuracy of the improved Bagging algorithm is greatly improved.

[0081] Step S14: Obtain the diagnostic dataset corresponding to the interior of the gas-insulated switchgear under test, and determine the target diagnostic result including fault type and fault confidence based on the diagnostic dataset and the target fault diagnosis model; the diagnostic dataset is the sulfur hexafluoride decomposition component corresponding to the interior of the gas-insulated switchgear under test.

[0082] In this embodiment, after obtaining the target fault diagnosis model, a diagnostic dataset is constructed using the decomposition components of sulfur hexafluoride (SF6), the insulating medium, when a fault occurs inside the gas-insulated switchgear under test. The target fault diagnosis model is then used to determine the target fusion result corresponding to the diagnostic dataset. Finally, the target fault type and corresponding fault confidence level are determined based on the basic probability assignment with the largest value in the target fusion result. Specifically, obtaining the diagnostic dataset corresponding to the inside of the gas-insulated switchgear under test, and determining the target diagnostic result including fault type and fault confidence level based on the diagnostic dataset and the target fault diagnosis model, includes: constructing a diagnostic dataset based on the decomposition components of SF6, the insulating medium, when a fault occurs inside the gas-insulated switchgear under test; determining the target fusion result corresponding to the diagnostic dataset using the target fault diagnosis model; and determining the target diagnostic result including fault type and fault confidence level based on the basic probability assignment with the largest value in the target fusion result.

[0083] As can be seen from the above, this application tests the performance of different numbers of base classifiers through pre-set test experiments and sulfur hexafluoride decomposition component datasets to determine the target number. This avoids computational redundancy caused by too many base classifiers and insufficient diversity caused by too few base classifiers. Based on the sulfur hexafluoride decomposition component dataset, sample sets are generated by sampling with replacement. Each sample set differs due to the randomness of sampling. Then, the corresponding base classifier is trained using the Bagging algorithm. The preliminary diagnostic results corresponding to the base classifiers are converted into basic probability assignments. The diagnostic confidence of the preliminary diagnostic results of the base classifiers is quantified to avoid misjudgment. High-conflict scenarios are identified based on the basic probability assignments, i.e., the preliminary diagnostic results of base classifier 1 are completely opposite to those of base classifier 2. The Dempster synthesis rule is improved by using the Dubois-Prade rule, which significantly improves the diagnostic accuracy of the target fault diagnosis model in high-conflict scenarios. In this way, by inputting the diagnostic dataset corresponding to the gas-insulated switchgear under test into the target fault diagnosis model, it is possible not only to accurately locate the fault type inside the gas-insulated switchgear under test, but also to ensure the reliability of the diagnostic results, improve the efficiency and generalization ability of insulation defect fault identification, and thus quickly investigate and deal with the fault, and maintain the safe and stable operation of the power system.

[0084] Accordingly, see Figure 9As shown, this application also provides a fault diagnosis device for gas-insulated switchgear, comprising:

[0085] The base classifier training module 11 is used to determine the target number of base classifiers using a preset test experiment and a sulfur hexafluoride decomposition component dataset, determine a sample set corresponding to the target number based on the sulfur hexafluoride decomposition component dataset and using sampling with replacement, train the base classifiers for the target number based on the sample set and the Bagging algorithm to obtain trained base classifiers, and input the sample test set into the trained base classifiers to obtain preliminary diagnostic results; the sulfur hexafluoride decomposition component dataset includes sulfur hexafluoride decomposition components and fault types.

[0086] The scene judgment module 12 is used to convert the preliminary diagnosis result into a basic probability assignment, and judge whether the current scene meets the preset high conflict conditions based on the basic probability assignment, so as to obtain the corresponding judgment result.

[0087] The diagnostic model determination module 13 is used to improve the Dempster synthesis rule based on the judgment result and using the Dubois-Prade rule to obtain the improved Dempster synthesis rule, and to fuse the basic probability assignment using the improved Dempster synthesis rule, and to determine the target fault diagnosis model based on the fusion result.

[0088] The diagnostic result determination module 14 is used to acquire the diagnostic dataset corresponding to the interior of the gas-insulated switchgear under test, and to determine the target diagnostic result, including the fault type and fault confidence, based on the diagnostic dataset and the target fault diagnosis model; the diagnostic dataset is the sulfur hexafluoride decomposition component corresponding to the interior of the gas-insulated switchgear under test.

[0089] In some specific embodiments, the base classifier training module 11 may specifically include:

[0090] The dataset partitioning unit is used to partition the sulfur hexafluoride decomposition component dataset according to a preset partitioning ratio to obtain a training set and a test set.

[0091] The integration result determination unit is used to construct several sample sets based on the training set and using sampling with replacement, train a corresponding number of base classifiers using the sample sets, and combine the output results of the base classifiers using the voting rule of the Bagging algorithm to obtain the integration result.

[0092] The label comparison unit is used to compare the integrated results with the first real fault label in the training set and the second real fault label in the test set, respectively, so as to determine the training set accuracy and the test set accuracy using the comparison results.

[0093] The target number determination unit is used to determine the number of base classifiers that meet the preset comprehensive performance conditions by using the training set accuracy and test set accuracy corresponding to different numbers of base classifiers, and to determine the number of base classifiers as the target number.

[0094] In some specific embodiments, the base classifier training module 11 may specifically include:

[0095] The dataset sampling unit is used to repeatedly sample the target quantity a number of times based on the sulfur hexafluoride decomposition component dataset using sampling with replacement, so as to obtain a sample set; the sample set has the same length as the sulfur hexafluoride decomposition component dataset.

[0096] The classifier training unit is used to train the base classifier based on the Bagging algorithm and using the sample set to obtain each trained base classifier, and input the sample test set into the trained base classifier to obtain each preliminary diagnostic result; the sample test set is a dataset determined based on the sulfur hexafluoride decomposition component dataset.

[0097] In some specific embodiments, the scene determination module 12 may specifically include:

[0098] A diagnostic result conversion unit is used to convert the preliminary diagnostic result into a basic probability assignment based on preset constraints; the preset constraints include that the confidence level of the empty set in the basic probability assignment is a first preset value, and that the total confidence level in the basic probability assignment is a second preset value.

[0099] A coefficient determination unit is used to determine the conflict coefficient between each of the basic probability assignments and to determine whether the conflict coefficient is not less than a first preset threshold.

[0100] The first scene determination unit is used to characterize the current scene as a scene that meets the preset high conflict condition if the conflict coefficient is not less than the first preset threshold.

[0101] In some specific embodiments, the scene determination module 12 may specifically include:

[0102] The criterion value judgment unit is used to determine the criterion value between each of the basic probability assignments and to determine whether the criterion value is not less than a second preset threshold.

[0103] The second scene determination unit is used to characterize the current scene as a scene that meets the preset high conflict condition if the criterion value is not less than the second preset threshold.

[0104] In some specific embodiments, the diagnostic model determination module 13 may specifically include:

[0105] The assignment fusion unit is used to improve the Dempster synthesis rule using the Dubois-Prade rule if the current scenario meets the preset high conflict conditions, so as to obtain the improved Dempster synthesis rule, and to fuse the basic probability assignment using the improved Dempster synthesis rule, and to determine the target fault diagnosis model based on the fusion result.

[0106] The output result determination unit is used to fuse the basic probability assignments using the Dempster synthesis rule if the current scenario does not meet the preset high conflict condition, and to determine the output result of the target fault diagnosis model based on the fusion result.

[0107] In some specific embodiments, the diagnostic result determination module 14 may specifically include:

[0108] The dataset construction unit is used to construct a diagnostic dataset based on the decomposition components of sulfur hexafluoride, the insulating medium, when a fault occurs inside the gas-insulated switchgear under test.

[0109] The target result determination unit is used to determine the target fusion result corresponding to the dataset to be diagnosed using the target fault diagnosis model, and to determine the target diagnosis result including fault type and fault confidence based on the basic probability assignment with the largest value in the target fusion result.

[0110] Furthermore, embodiments of this application also disclose an electronic device, Figure 10 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the fault diagnosis method for gas-insulated combined electrical appliances disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0111] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0112] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0113] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the fault diagnosis method for the gas-insulated combined electrical appliance disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0114] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned fault diagnosis method for gas-insulated switchgear. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0115] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0116] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0117] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0118] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0119] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of diagnosing a fault of a gas-insulated combined electrical apparatus, characterized by, The method comprises the following steps: determining a target number of base classifiers by using a preset test experiment and a sulfur hexafluoride decomposition component dataset, determining a sample set corresponding to the target number based on the sulfur hexafluoride decomposition component dataset and by using a sampling method with replacement, training the base classifiers of the target number based on the sample set and a Bagging algorithm to obtain trained base classifiers, inputting a sample test set into the trained base classifiers to obtain preliminary diagnosis results, and the sulfur hexafluoride decomposition component dataset comprises sulfur hexafluoride decomposition components and fault types; converting the preliminary diagnosis results into basic probability assignments, judging whether a current scene is a scene satisfying a preset high conflict condition based on the basic probability assignments to obtain a corresponding judgment result; improving a Dempster combination rule based on the judgment result and by using a Dubois-Prade rule to obtain an improved Dempster combination rule, fusing the basic probability assignments by using the improved Dempster combination rule, and determining a target fault diagnosis model based on a fusion result; obtaining a to-be-diagnosed dataset corresponding to an internal part of a to-be-tested gas insulated switchgear, determining a target diagnosis result comprising a fault type and a fault confidence by using the target fault diagnosis model based on the to-be-diagnosed dataset, and the to-be-diagnosed dataset is a sulfur hexafluoride decomposition component corresponding to the internal part of the to-be-tested gas insulated switchgear.

2. The method of diagnosing a failure of a gas insulated combined electrical apparatus according to claim 1, characterized by, The method of determining the target number of base classifiers by using the preset test experiment and the sulfur hexafluoride decomposition component dataset comprises the following steps: dividing the sulfur hexafluoride decomposition component dataset based on a preset division ratio to obtain a training set and a test set; constructing a plurality of sample sets by using a sampling method with replacement based on the training set, training a corresponding number of base classifiers by using the sample sets, and comprehensively integrating output results of the base classifiers by using a voting rule of a Bagging algorithm to obtain an integrated result; comparing the integrated result with a first real fault label in the training set and a second real fault label in the test set respectively to determine a training set accuracy and a test set accuracy by using comparison results; determining a number of base classifiers satisfying a preset comprehensive performance condition by using the training set accuracy and the test set accuracy of the base classifiers corresponding to different numbers, and determining the number of base classifiers as the target number.

3. The method of diagnosing a failure of a gas insulated combined electrical apparatus according to claim 1, characterized by, The method of determining the target number of base classifiers by using the preset test experiment and the sulfur hexafluoride decomposition component dataset comprises the following steps: repeatedly sampling a number of times corresponding to the target number by using a sampling method with replacement based on the sulfur hexafluoride decomposition component dataset to obtain a sample set, and the sample set has the same length as the sulfur hexafluoride decomposition component dataset; training the base classifier based on the bagging algorithm and using the sample set to obtain each trained base classifier, and inputting a sample test set into the trained base classifier to obtain each preliminary diagnosis result; the sample test set is a data set determined based on the sulfur hexafluoride decomposition component data set.

4. The method of diagnosing a failure of a gas insulated combined electrical apparatus according to claim 1, characterized by, The preliminary diagnosis result is converted into a basic probability assignment based on the basic probability assignment to determine whether the current scene is a scene that meets the preset high conflict condition to obtain a corresponding judgment result, including: The preliminary diagnosis result is converted into a basic probability assignment based on the preset constraint condition; the preset constraint condition includes that the empty set confidence in the basic probability assignment is a first preset value, and the total confidence in the basic probability assignment is a second preset value; Determine the conflict coefficient between each basic probability assignment, and determine whether the conflict coefficient is not less than a first preset threshold; If the conflict coefficient is not less than the first preset threshold, it indicates that the current scene is a scene that meets the preset high conflict condition.

5. The method of diagnosing a failure of a gas insulated combined electrical apparatus according to claim 4, characterized by, After the preliminary diagnosis result is converted into a basic probability assignment based on the preset constraint condition, it further includes: Determine the criterion value between each basic probability assignment, and determine whether the criterion value is not less than a second preset threshold; If the criterion value is not less than the second preset threshold, it indicates that the current scene is a scene that meets the preset high conflict condition.

6. The method of diagnosing a failure of a gas insulated combined electrical apparatus according to claim 1, characterized by, The judgment result is used to improve the Dempster combination rule based on the Dubois-Prade rule to obtain an improved Dempster combination rule, and the basic probability assignment is fused based on the improved Dempster combination rule, and a target fault diagnosis model is determined based on the fusion result, including: If the current scene is a scene that meets the preset high conflict condition, the Dempster combination rule is improved based on the Dubois-Prade rule to obtain an improved Dempster combination rule, and the basic probability assignment is fused based on the improved Dempster combination rule, and a target fault diagnosis model is determined based on the fusion result; If the current scene is a scene that does not meet the preset high conflict condition, the basic probability assignment is fused based on the Dempster combination rule, and the output result of the target fault diagnosis model is determined based on the fusion result.

7. The method of diagnosing a fault of a gas-insulated combined electrical apparatus according to any one of claims 1 to 6, characterized by, The target diagnosis result including fault type and fault confidence is determined based on the target fault diagnosis model and the to-be-diagnosed data set corresponding to the to-be-tested gas-insulated combined electric appliance, including: Based on the decomposition components of the insulating medium sulfur hexafluoride corresponding to the to-be-diagnosed data set when the to-be-tested gas-insulated combined electric appliance fails; The target fusion result corresponding to the to-be-diagnosed data set is determined using the target fault diagnosis model, and the target diagnosis result including fault type and fault confidence is determined based on the basic probability assignment with the maximum value in the target fusion result.

8. A fault diagnosis device for gas-insulated combined electrical appliances, characterized in that, including: The base classifier training module is configured to determine a target number of base classifiers by using a preset test experiment and a sulfur hexafluoride decomposition component dataset, determine a sample set corresponding to the target number by using a sampling method with replacement based on the sulfur hexafluoride decomposition component dataset, train the base classifiers of the target number based on the sample set and a Bagging algorithm to obtain trained base classifiers, input a sample test set into the trained base classifiers to obtain preliminary diagnosis results, and the sulfur hexafluoride decomposition component dataset includes sulfur hexafluoride decomposition components and fault types. The scene judgment module is configured to convert the preliminary diagnosis results into basic probability assignments, judge whether a current scene is a scene meeting a preset high conflict condition based on the basic probability assignments, and obtain corresponding judgment results. The diagnosis model determination module is configured to improve a Dempster combination rule by using a Dubois-Prade rule based on the judgment results to obtain an improved Dempster combination rule, fuse the basic probability assignments by using the improved Dempster combination rule, and determine a target fault diagnosis model based on a fusion result. The diagnosis result determination module is configured to obtain a to-be-diagnosed dataset corresponding to an internal part of a to-be-tested gas insulated switchgear, determine a target diagnosis result including a fault type and a fault confidence level based on the to-be-diagnosed dataset and by using the target fault diagnosis model, and the to-be-diagnosed dataset is a sulfur hexafluoride decomposition component corresponding to the internal part of the to-be-tested gas insulated switchgear.

9. An electronic device, comprising: The computer program is stored in the memory and executed by the processor to implement the fault diagnosis method of the gas insulated switchgear. The computer program is stored in the memory and executed by the processor to implement the fault diagnosis method of the gas insulated switchgear. The computer program is stored in the memory and executed by the processor to implement the fault diagnosis method of the gas insulated switchgear.

10. A computer-readable storage medium, characterized in that, ​

Images