Transformer fault detection method, device, equipment, medium and program product

Abstract

The invention provides a transformer fault detection method, device and equipment, a medium and a program product. In the method, according to acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duwal triangle data and model detection accuracy data, a first fuzzy prediction model, a second fuzzy prediction model and a third fuzzy prediction model, the dissolved gas concentration data, the dissolved gas ratio data, the dissolved gas Duwal triangle data and the model detection accuracy data are acquired; and determining a first fault type probability vector, a second fault type probability vector and a third fault type probability vector. And inputting the first fault type probability vector, the second fault type probability vector and the third fault type probability vector into a classification model to obtain a detection result which is a fault-free or target fault type. According to the scheme, fault detection is carried out on the transformer through the dissolved gas concentration data, the dissolved gas ratio data, the dissolved gas Duwal triangle data and the model detection accuracy data, and the detection accuracy is effectively improved.

Description

Technical Field

[0001] This application relates to the field of transformer technology, and in particular to a transformer fault detection method, device, equipment, medium, and program product. Background Technology

[0002] As power grids grow larger, the operational safety and stability of power systems have become a major concern. Transformers are a crucial component of the power system; transformer failures can negatively impact the system's operational safety and stability, making transformer fault detection essential.

[0003] In existing technologies, transformer fault detection typically involves measuring the concentration of dissolved gases in the transformer oil and then determining whether the transformer has malfunctioned based on whether the gas concentration exceeds a preset threshold.

[0004] In summary, existing transformer fault detection methods rely solely on gas concentration for fault detection, resulting in low accuracy. Summary of the Invention

[0005] The transformer fault detection method, apparatus, equipment, medium, and program products provided in this application are intended to solve the problem that existing transformer fault detection methods rely solely on gas concentration for transformer fault detection, resulting in low detection accuracy.

[0006] In a first aspect, embodiments of this application provide a transformer fault detection method, including:

[0007] Based on the obtained dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangle data, and model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model, the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model are determined.

[0008] The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector are input into the classification model to obtain the detection result, which is either no fault or the target fault type.

[0009] The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector all include the fault probability of each fault type.

[0010] In one possible implementation, the dissolved gas concentration data includes: hydrogen concentration, methane concentration, and carbon monoxide concentration;

[0011] The dissolved gas ratio data includes: the ratio of methane concentration to hydrogen concentration, and the ratio of acetylene concentration to ethylene concentration;

[0012] The dissolved gas Duval triangle data includes: the ratio of methane concentration to the overall gas concentration, the ratio of acetylene concentration to the overall gas concentration, and the ratio of ethylene concentration to the overall gas concentration; the overall gas concentration is the sum of the methane concentration, the acetylene concentration, and the ethylene concentration.

[0013] The model detection accuracy data includes: the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type.

[0014] In one possible implementation, the step of determining the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model based on the acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangular data, and model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model, includes:

[0015] Using the first fuzzy prediction model, the dissolved gas concentration data, the dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type are predicted to obtain the first fault type probability vector.

[0016] Using the second fuzzy prediction model, the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type are predicted to obtain the second fault type probability vector;

[0017] Using the third fuzzy prediction model, the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type are predicted to obtain the probability vector of the third fault type.

[0018] In one possible implementation, the step of using the first fuzzy prediction model to predict the dissolved gas concentration data, the dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type to obtain the first fault type probability vector includes:

[0019] The following processing is performed using the first fuzzy prediction model:

[0020] Based on the dissolved gas concentration data, calculate the membership degree of each data point in the dissolved gas concentration data at each preset concentration level;

[0021] Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data under each preset ratio level;

[0022] Based on the membership degree of each data point in the dissolved gas concentration data at each preset concentration level, and the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, the matching degree of each fault type is determined;

[0023] For each fault type, the product of the matching degree of the fault type and the first accuracy of the first fuzzy prediction model for the fault type is taken as the fault probability of the fault type.

[0024] Generate the first fault type probability vector based on the fault probability of each fault type.

[0025] In one possible implementation, the step of employing the second fuzzy prediction model to predict the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type, thereby obtaining the second fault type probability vector, includes:

[0026] The second fuzzy prediction model is used for the following processing:

[0027] Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data under each preset ratio level;

[0028] Based on the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, the matching degree of each fault type is determined;

[0029] For each fault type, the product of the matching degree of the fault type and the second accuracy of the second fuzzy prediction model for the fault type is taken as the fault probability of the fault type.

[0030] Based on the failure probability of each failure type, a second failure type probability vector is generated.

[0031] In one possible implementation, the step of employing the third fuzzy prediction model to predict the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type, thereby obtaining the third fault type probability vector, includes:

[0032] The third fuzzy prediction model is used for the following processing:

[0033] Based on the dissolved gas Duval triangular data, calculate the membership degree of each data point in the dissolved gas Duval triangular data at each preset percentage level;

[0034] Based on the membership degree of each data point in the dissolved gas Duval triangle data at each preset percentage level, the matching degree of each fault type is determined;

[0035] For each fault type, the product of the matching degree of the fault type and the third accuracy of the third fuzzy prediction model for the fault type is taken as the fault probability of the fault type.

[0036] The third fault type probability vector is generated based on the fault probability of each fault type.

[0037] In one possible implementation, the method further includes:

[0038] If the detection result is the target fault type, then the fusion detection accuracy of each fault type is calculated based on the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type.

[0039] Output the fusion detection accuracy of the target fault type.

[0040] Secondly, embodiments of this application provide a transformer fault detection device, comprising:

[0041] The processing module is used to determine the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model based on the acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangular data, model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model.

[0042] The detection module is used to input the first fault type probability vector, the second fault type probability vector and the third fault type probability vector into the classification model to obtain the detection result, wherein the detection result is no fault or the target fault type;

[0043] The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector all include the fault probability of each fault type.

[0044] Thirdly, embodiments of this application provide a server, including:

[0045] Processor, memory, communication interface;

[0046] The memory is used to store the executable instructions of the processor;

[0047] The processor is configured to execute the transformer fault detection method according to any one of the first aspects by executing the executable instructions.

[0048] Fourthly, embodiments of this application provide a readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the transformer fault detection method described in any of the first aspects.

[0049] Fifthly, embodiments of this application provide a computer program product, including a computer program, which, when executed by a processor, is used to implement the transformer fault detection method described in any of the first aspects.

[0050] The transformer fault detection method, apparatus, equipment, medium, and program products provided in this application, based on acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangulation data, and model detection accuracy data, as well as a first fuzzy prediction model, a second fuzzy prediction model, and a third fuzzy prediction model, determine the probability vectors for the first, second, and third fault types corresponding to the first, second, and third fuzzy prediction models, respectively. These probability vectors are then input into a classification model to obtain the detection result, which is either no fault or the target fault type. This solution effectively improves detection accuracy by using dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangulation data, and model detection accuracy data to detect transformer faults. Attached Figure Description

[0051] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0052] Figure 1 A flowchart illustrating an embodiment of the transformer fault detection method provided in this application;

[0053] Figure 2 A flowchart illustrating Embodiment 2 of the transformer fault detection method provided in this application;

[0054] Figure 3 A flowchart illustrating Embodiment 3 of the transformer fault detection method provided in this application;

[0055] Figure 4A flowchart illustrating Embodiment 4 of the transformer fault detection method provided in this application;

[0056] Figure 5 A flowchart illustrating Embodiment 5 of the transformer fault detection method provided in this application;

[0057] Figure 6 A flowchart illustrating Embodiment Six of the transformer fault detection method provided in this application;

[0058] Figure 7 This is a schematic diagram of the structure of an embodiment of the transformer fault detection device provided in this application;

[0059] Figure 8 This is a schematic diagram of the structure of an electronic device provided in this application.

[0060] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0061] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0062] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0063] As power grids grow larger, the operational safety and stability of power systems have become a major concern. Transformers are a crucial component of the power system; transformer failures can negatively impact the system's operational safety and stability, making transformer fault detection essential.

[0064] In existing technologies, transformer fault detection typically involves measuring the concentration of dissolved gases in the transformer oil and then determining whether a fault has occurred based on whether the gas concentration exceeds a preset threshold. For example, a partial discharge fault is identified when the hydrogen concentration exceeds 150 ppm. However, relying solely on gas concentration for transformer fault detection results in low accuracy.

[0065] To address the problems existing in the prior art, the inventors, during their research on transformer fault detection methods, discovered that transformer faults can be categorized into various types, such as partial discharge faults, thermal faults, and arcing faults. Different fault types not only result in varying concentrations of dissolved gases in the transformer oil but also different dissolved gas ratios and Duval trigonometric identities. Therefore, by combining dissolved gas concentration data, dissolved gas ratio data, and dissolved gas Duval trigonometric identity data with model detection accuracy data and fuzzy prediction models, transformer fault detection can be performed, effectively improving detection accuracy. Based on the above inventive concept, the transformer fault detection scheme described in this application was designed.

[0066] The transformer fault detection method in this application can be executed by a computer, a server, a terminal device, etc. This application does not limit it. The following description uses a computer as an example.

[0067] The following provides examples illustrating the application scenarios of the transformer fault detection method provided in this application.

[0068] For example, in this application scenario, to ensure the safety and stability of the power grid, fault detection of transformers is required. A gas sensor is installed in the transformer oil, and the gas sensor transmits the collected dissolved gas concentration data to a computer for fault detection. The computer also stores model detection accuracy data, a first fuzzy prediction model, a second fuzzy prediction model, and a third fuzzy prediction model.

[0069] The computer calculates dissolved gas ratio data and dissolved gas Duval triangle data based on dissolved gas concentration data.

[0070] Based on dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangle data, and model detection accuracy data, as well as the first, second, and third fuzzy prediction models, the computer determines the probability vectors for the first, second, and third fault types corresponding to the first, second, and third fuzzy prediction models, respectively. Each of these probability vectors includes the probability of each fault type.

[0071] To further improve detection accuracy, the computer inputs the probability vectors of the first, second, and third fault types into the classification model to obtain the detection results, which are either no fault or the target fault type.

[0072] The computer can then display the test results, the probability vector of the first fault type, the probability vector of the second fault type, the probability vector of the third fault type, and dissolved gas concentration data, so that users can view and perform timely repairs.

[0073] It should be noted that the above scenario is only an example of an application scenario provided by the embodiments of this application. The embodiments of this application do not limit the actual form of the various devices included in the scenario, nor do they limit the interaction method between devices. In the specific application of the solution, it can be set according to actual needs.

[0074] The technical solution of this application will now be described in detail through specific embodiments. It should be noted that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0075] Figure 1 This is a flowchart illustrating an embodiment of the transformer fault detection method provided in this application. This embodiment describes how a computer performs transformer fault detection based on dissolved gas concentration data, dissolved gas ratio data, and dissolved gas Duval triangle data. The method in this embodiment can be implemented through software, hardware, or a combination of both. Figure 1 As shown, the transformer fault detection method specifically includes the following steps:

[0076] S101: Based on the acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangulation data, and model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model, determine the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model.

[0077] In this step, in order to detect faults in the transformer, the computer needs to acquire dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangulation data, and model detection accuracy data. Then, based on these data and the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model, the computer determines the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model.

[0078] The dissolved gas concentration data includes: hydrogen concentration, methane concentration, and carbon monoxide concentration.

[0079] It should be noted that the concentrations of hydrogen, methane, and carbon monoxide can be the original concentrations of the gases in the transformer oil, or the concentrations after normalization of the original concentrations to eliminate dimensional differences.

[0080] Formulas can be used Calculate the normalized concentration. Among them, This represents the normalized concentration. Indicates the original concentration. Indicates the lowest concentration of the gas. This indicates the highest concentration of the gas.

[0081] For hydrogen, It can be 90ppm, 100ppm, 110ppm, etc. It can be 900ppm, 1000ppm, 1100ppm, etc. For methane, It can be 95ppm, 100ppm, 115ppm, etc. It can be 910 ppm, 1000 ppm, 1150 ppm, etc. For carbon monoxide, It can be 85ppm, 100ppm, 105ppm, etc. It can be 950ppm, 1000ppm, 1050ppm, etc. The embodiments in this application are not... and Limitations can be set based on the actual situation.

[0082] Dissolved gas ratio data include: the ratio of methane concentration to hydrogen concentration, and the ratio of acetylene concentration to ethylene concentration.

[0083] It should be noted that the computer can obtain dissolved gas ratio data in several ways: First, the computer can obtain the concentrations of methane, hydrogen, acetylene, and ethylene through gas sensors, and then calculate the ratios of methane to hydrogen concentration and acetylene to ethylene concentration to obtain the dissolved gas ratio data. Second, the user can input the dissolved gas ratio data into the computer, which can then obtain the data. Third, the user can send the dissolved gas ratio data to the computer using a terminal device, which can then obtain the data. This application does not limit the method by which the computer obtains the dissolved gas ratio data; it can be determined according to the actual situation.

[0084] The dissolved gas Duval triangle data includes: the ratio of methane concentration to the total gas concentration, the ratio of acetylene concentration to the total gas concentration, and the ratio of ethylene concentration to the total gas concentration. The total gas concentration is the sum of the concentrations of methane, acetylene, and ethylene.

[0085] It should be noted that the computer can obtain the dissolved gas Duval triangulation data in several ways: First, the computer obtains the concentrations of methane, acetylene, and ethylene through gas sensors, then calculates the overall gas concentration, and finally calculates the ratios of methane concentration to the overall gas concentration, acetylene concentration to the overall gas concentration, and ethylene concentration to the overall gas concentration to obtain the dissolved gas Duval triangulation data. Second, the user can input the dissolved gas Duval triangulation data into the computer, which can then obtain the data. Third, the user can send the dissolved gas Duval triangulation data to the computer using a terminal device, which can then obtain the data. This application does not limit the method by which the computer obtains the dissolved gas Duval triangulation data; it can be determined according to the actual situation.

[0086] The model detection accuracy data includes: the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type.

[0087] The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector all include the fault probability of each fault type.

[0088] It should be noted that the fault type can be partial discharge fault, thermal fault, arc fault, and arc-thermal mixed fault, etc. The embodiments of this application do not limit the fault type, and it can be determined according to the actual situation.

[0089] Specifically, a first fuzzy prediction model is used to predict the dissolved gas concentration data, dissolved gas ratio data, and the first accuracy rate of the first fuzzy prediction model for each fault type, thereby obtaining a first fault type probability vector. In other words, the dissolved gas concentration data, dissolved gas ratio data, and the first accuracy rate of the first fuzzy prediction model for each fault type are input into the first fuzzy prediction model to obtain the first fault type probability vector.

[0090] A second fuzzy prediction model is used to predict the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type, resulting in a second fault type probability vector. In other words, the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type are input into the second fuzzy prediction model to obtain the second fault type probability vector.

[0091] A third fuzzy prediction model is used to predict the third fault type probability vector by processing the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type. In other words, the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type are input into the third fuzzy prediction model to obtain the third fault type probability vector.

[0092] Because different fault types not only result in different concentrations of dissolved gases in transformer oil, but also different dissolved gas ratios and different dissolved gas Duval triangulation data, the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model can be used to obtain the fault type probability vectors respectively, so as to improve the detection accuracy.

[0093] S102: Input the probability vectors of the first fault type, the second fault type, and the third fault type into the classification model to obtain the detection results.

[0094] In this step, after the computer obtains the first fault type probability vector, the second fault type probability vector, and the third fault type probability vector, in order to determine whether the transformer has a fault, and if a fault occurs, to determine the type of fault, the first fault type probability vector, the second fault type probability vector, and the third fault type probability vector are input into the classification model to obtain the detection result, which is either no fault or the target fault type.

[0095] It should be noted that the target fault type can be partial discharge fault, thermal fault, arc fault, and arc-thermal mixed fault, etc. The embodiments of this application do not limit the target fault type, and it can be determined according to the actual situation.

[0096] It should be noted that the classification model can be a Support Vector Machine (SVM) model, a neural network model, a random forest model, a Naive Bayes model, etc. This application does not limit the classification model, and it can be determined according to the actual situation.

[0097] The computer can then display the test results, the probability vector of the first fault type, the probability vector of the second fault type, the probability vector of the third fault type, and dissolved gas concentration data, so that users can view and perform timely repairs.

[0098] The transformer fault detection method provided in this embodiment determines the probability vectors for the first fault type (first fuzzy prediction model), the second fault type (second fuzzy prediction model), and the third fault type (third fuzzy prediction model) based on acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangulation data, and model detection accuracy data, as well as a first fuzzy prediction model, a second fuzzy prediction model, and a third fuzzy prediction model. These probability vectors are then input into a classification model to obtain the detection result, which is either no fault or the target fault type. Compared to existing technologies that only rely on gas concentration for transformer fault detection, this method utilizes dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangulation data, and model detection accuracy data to detect transformer faults, effectively improving detection accuracy and efficiency. Furthermore, the combination of the three fault type probability vectors obtained from the three fuzzy prediction models with the classification model further enhances detection accuracy.

[0099] Figure 2 This is a flowchart illustrating a second embodiment of the transformer fault detection method provided in this application. Based on the above embodiments, this application describes how a computer uses a first fuzzy prediction model to predict dissolved gas concentration data, dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type, thereby obtaining a probability vector for the first fault type. For example... Figure 2 As shown, the following steps are performed using the first fuzzy prediction model:

[0100] S201: Based on the dissolved gas concentration data, calculate the membership degree of each data point in the dissolved gas concentration data at each preset concentration level.

[0101] In this step, after the computer obtains the dissolved gas concentration data, dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type, in order to obtain the probability vector of the first fault type, it is necessary to calculate the membership degree of each data point in the dissolved gas concentration data at each preset concentration level based on the dissolved gas concentration data.

[0102] For example, the preset concentration level is low, medium, or high. This can be determined according to the formula. Calculate the membership degree of hydrogen concentration at low levels, where, This indicates the degree of membership of hydrogen concentration at lower levels. Indicates hydrogen concentration. This indicates the preset center value of hydrogen concentration at low levels. This indicates the preset concentration range of hydrogen at low levels.

[0103] According to the formula Calculate the membership degree of hydrogen concentration at the medium level, where, This indicates the membership degree of hydrogen concentration at the intermediate level. Indicates hydrogen concentration. This indicates the preset center value of hydrogen concentration at the intermediate level. This indicates the preset concentration range of hydrogen at the intermediate level.

[0104] According to the formula Calculate the membership degree of hydrogen concentration at higher levels, where, This indicates the degree of membership of hydrogen concentration at higher levels. Indicates hydrogen concentration. This indicates the preset center value of hydrogen concentration at a high level. This indicates the preset concentration range of hydrogen at a high level.

[0105] It should be noted that It can be 240ppm, 250ppm, 260ppm, etc. It could be 540ppm, 550ppm, 560ppm, etc. It could be 840ppm, 850ppm, 860ppm, etc. , and It can be 280ppm, 300ppm, 320ppm, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0106] According to the formula Calculate the membership degree of methane concentration at lower levels, where, This indicates the membership degree of methane concentration at lower levels. Indicates methane concentration. This indicates the preset center value of methane concentration at low levels. This indicates the preset concentration range of methane at low levels.

[0107] According to the formula Calculate the membership degree of methane concentration at the intermediate level, where, This indicates the membership degree of methane concentration at the intermediate level. Indicates methane concentration. This indicates the preset center value of methane concentration at the intermediate level. This indicates the preset concentration range scale for methane at the medium level.

[0108] According to the formula Calculate the membership degree of methane concentration at higher levels, where, This indicates the membership degree of methane concentration at higher levels. Indicates methane concentration. This indicates the preset center value of methane concentration at a high level. This indicates the preset concentration range of methane at high levels.

[0109] It should be noted that It can be 240ppm, 250ppm, 260ppm, etc. It could be 540ppm, 550ppm, 560ppm, etc. It could be 840ppm, 850ppm, 860ppm, etc. , and It can be 280ppm, 300ppm, 320ppm, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0110] According to the formula Calculate the membership degree of carbon monoxide concentration at low levels, where, This indicates the degree of membership of carbon monoxide concentration at lower levels. Indicates carbon monoxide concentration. This indicates the preset center value of carbon monoxide concentration at a low level. This indicates the preset concentration range of carbon monoxide at low levels.

[0111] According to the formula Calculate the membership degree of carbon monoxide concentration at the medium level, where, This indicates the degree of membership of carbon monoxide concentration at the medium level. Indicates carbon monoxide concentration. This indicates the preset center value of carbon monoxide concentration at the intermediate level. This indicates the preset concentration range of carbon monoxide at the intermediate level.

[0112] According to the formula Calculate the membership degree of carbon monoxide concentration at higher levels, where, This indicates the degree of membership of carbon monoxide concentration at higher levels. Indicates carbon monoxide concentration. This indicates the preset center value of carbon monoxide concentration at a high level. This indicates the preset concentration range of carbon monoxide at high levels.

[0113] It should be noted that It can be 240ppm, 250ppm, 260ppm, etc. It could be 540ppm, 550ppm, 560ppm, etc. It could be 840ppm, 850ppm, 860ppm, etc. , and It can be 280ppm, 300ppm, 320ppm, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0114] S202: Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data at each preset ratio level.

[0115] In this step, after the computer obtains the dissolved gas concentration data, dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type, in order to obtain the probability vector of the first fault type, it is necessary to calculate the membership degree of each data in the dissolved gas ratio data under each preset ratio level based on the dissolved gas ratio data.

[0116] For example, the preset ratio levels are low, medium, or high. This can be determined according to the formula. Calculate the membership degree of the ratio of methane concentration to hydrogen concentration at the smallest level, where, This indicates the membership degree of the ratio of methane concentration to hydrogen concentration at a smaller level. This represents the ratio of methane concentration to hydrogen concentration. This represents the preset center value of the ratio of methane concentration to hydrogen concentration at the lowest possible level. This indicates a preset range of ratios for methane to hydrogen concentrations at a small level.

[0117] According to the formula Calculate the membership degree of the ratio of methane concentration to hydrogen concentration at the medium level, where, This indicates the membership degree of the ratio of methane concentration to hydrogen concentration at the medium level. This represents the ratio of methane concentration to hydrogen concentration. This represents the preset center value of the ratio of methane concentration to hydrogen concentration at the medium level. This indicates a preset range of ratios for methane to hydrogen concentrations at the medium level.

[0118] According to the formula Calculate the membership degree of the ratio of methane concentration to hydrogen concentration at major levels, where, This represents the membership degree of the ratio of methane concentration to hydrogen concentration at higher levels. This represents the ratio of methane concentration to hydrogen concentration. This represents the preset center value of the ratio of methane concentration to hydrogen concentration at high levels. This indicates a preset range of ratios for methane to hydrogen concentrations at higher levels.

[0119] It should be noted that It can be 1.6, 1.7, 1.8, etc. It could be 4.9, 5, 5.1, etc. It could be 8.28.3, 8.4, etc. , and It can be 3.2, 3.3, 3.4, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0120] According to the formula Calculate the membership degree of the ratio of acetylene concentration to ethylene concentration at the smallest level, where, This indicates the membership degree of the ratio of acetylene concentration to ethylene concentration at the smallest level. This represents the ratio of acetylene concentration to ethylene concentration. This represents the preset center value of the ratio of acetylene concentration to ethylene concentration at the lowest possible level. This indicates a preset range of the ratio of acetylene concentration to ethylene concentration at a small level.

[0121] According to the formula Calculate the membership degree of the ratio of acetylene concentration to ethylene concentration at the medium level, where, This indicates the membership degree of the ratio of acetylene concentration to ethylene concentration at the medium level. This represents the ratio of acetylene concentration to ethylene concentration. This represents the preset center value of the acetylene-to-ethylene ratio at the medium level. This indicates a preset range of ratios for the acetylene concentration to the ethylene concentration at the medium level.

[0122] According to the formula Calculate the membership degree of the ratio of acetylene concentration to ethylene concentration at major levels, where, This indicates the membership degree of the ratio of acetylene concentration to ethylene concentration at higher levels. This represents the ratio of acetylene concentration to ethylene concentration. This represents the preset center value of the ratio of acetylene concentration to ethylene concentration at major levels. This indicates a preset range of ratios for the acetylene concentration to the ethylene concentration at higher levels.

[0123] It should be noted that It can be 1.6, 1.7, 1.8, etc. It could be 4.9, 5, 5.1, etc. It could be 8.28.3, 8.4, etc. , and It can be 3.2, 3.3, 3.4, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0124] It should be noted that the execution order of steps S201 and S202 can be: step S201 is executed first, then step S202. Alternatively, step S202 is executed first, then step S201. Or, steps S201 and S202 are executed simultaneously. This embodiment does not limit the execution order of steps S201 and S202; it can be determined according to the actual situation.

[0125] S203: Determine the matching degree of each fault type based on the membership degree of each data point in the dissolved gas concentration data at each preset concentration level and the membership degree of each data point in the dissolved gas ratio data at each preset ratio level.

[0126] In this step, after the computer obtains the membership degree of each data point in the dissolved gas concentration data at each preset concentration level and the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, it can determine the matching degree of each fault type based on these membership degrees.

[0127] Specifically, there is a corresponding relationship between each fault type and its membership degree. Therefore, for each fault type, the matching degree of the fault type can be determined based on the membership degree corresponding to that fault type.

[0128] The maximum, minimum, or average membership degree corresponding to the fault type can be used as the matching degree of the fault type.

[0129] For example, the membership degree corresponding to a partial discharge fault includes: the membership degree of hydrogen concentration at a high level, and the membership degree of the ratio of acetylene concentration to ethylene concentration at a low level.

[0130] The membership degrees corresponding to thermal faults include: the membership degree of methane concentration at medium levels, and the membership degree of the ratio of acetylene concentration to ethylene concentration at high levels.

[0131] The membership degree corresponding to the arc fault includes: the membership degree of carbon monoxide concentration at high levels, and the membership degree of the ratio of methane concentration to hydrogen concentration at low levels.

[0132] S204: For each fault type, the product of the matching degree of the fault type and the first accuracy of the first fuzzy prediction model for the fault type is used as the fault probability of the fault type.

[0133] In this step, after the computer obtains the matching degree for each fault type, in order to improve the detection accuracy, for each fault type, the product of the matching degree of the fault type and the first accuracy of the first fuzzy prediction model for that fault type is used as the fault probability of that fault type.

[0134] It should be noted that the matching degree of each fault type can be normalized before calculating the fault probability of each fault type.

[0135] For example, the first fuzzy prediction model has a first accuracy rate of 95.65% for partial discharge faults, a first accuracy rate of 91.391% for thermal faults, and a first accuracy rate of 84.61% for arc faults. This application does not limit the first accuracy rate of the first fuzzy prediction model for each fault type.

[0136] S205: Generate the first fault type probability vector based on the fault probability of each fault type.

[0137] In this step, after the computer obtains the failure probability of each failure type, it combines the failure probabilities of each failure type to obtain the first failure type probability vector.

[0138] The transformer fault detection method provided in this embodiment uses a first fuzzy prediction model to predict the dissolved gas concentration data, dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type, thereby obtaining a first fault type probability vector, which can improve the accuracy of the first fault type probability vector.

[0139] Figure 3 This is a flowchart illustrating Embodiment 3 of the transformer fault detection method provided in this application. Based on the above embodiments, this embodiment describes how the computer uses a second fuzzy prediction model to predict the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type, thereby obtaining a second fault type probability vector. For example... Figure 3 As shown, the following steps are performed using the second fuzzy prediction model:

[0140] S301: Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data at each preset ratio level.

[0141] It should be noted that this step is similar to step S202 in Embodiment 2, and will not be described again here.

[0142] S302: Determine the matching degree of each fault type based on the membership degree of each data point in the dissolved gas ratio data at each preset ratio level.

[0143] In this step, after the computer obtains the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, it can determine the matching degree of each fault type based on these membership degrees.

[0144] Specifically, there is a corresponding relationship between each fault type and its membership degree. Therefore, for each fault type, the matching degree of the fault type can be determined based on the membership degree corresponding to that fault type.

[0145] The maximum, minimum, or average membership degree corresponding to the fault type can be used as the matching degree of the fault type.

[0146] For example, the membership degree corresponding to a partial discharge fault includes: the membership degree of the ratio of acetylene concentration to ethylene concentration at the minor level, and the membership degree of the ratio of methane concentration to hydrogen concentration at the minor level.

[0147] The membership degrees corresponding to thermal faults include: the membership degree of the ratio of acetylene concentration to ethylene concentration at the lower level, and the membership degree of the ratio of methane concentration to hydrogen concentration at the higher level.

[0148] The membership degrees corresponding to arc faults include: the membership degree of the ratio of acetylene concentration to ethylene concentration at the higher level, and the membership degree of the ratio of methane concentration to hydrogen concentration at the lower level.

[0149] S303: For each fault type, the product of the matching degree of the fault type and the second accuracy of the second fuzzy prediction model for the fault type is used as the fault probability of the fault type.

[0150] In this step, after the computer obtains the matching degree for each fault type, in order to improve the detection accuracy, for each fault type, the product of the matching degree of the fault type and the second accuracy of the second fuzzy prediction model for that fault type is used as the fault probability of that fault type.

[0151] It should be noted that the matching degree of each fault type can be normalized before calculating the fault probability of each fault type.

[0152] For example, the second fuzzy prediction model has a second accuracy rate of 78.26% for partial discharge faults, a second accuracy rate of 97.351% for thermal faults, and a second accuracy rate of 88.81% for arc faults. This application does not limit the second accuracy rate of the second fuzzy prediction model for each fault type.

[0153] S304: Generate a second fault type probability vector based on the fault probability of each fault type.

[0154] In this step, after the computer obtains the failure probability of each failure type, it combines the failure probabilities of each failure type to obtain the second failure type probability vector.

[0155] The transformer fault detection method provided in this embodiment improves the accuracy of the second fault type probability vector by using a second fuzzy prediction model on the computer to predict the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type.

[0156] Figure 4 This is a flowchart illustrating Embodiment 4 of the transformer fault detection method provided in this application. Based on the above embodiments, this embodiment describes how the computer uses a third fuzzy prediction model to predict the third accuracy of the dissolved gas Duval triangulation data and the third fuzzy prediction model for each fault type, obtaining a third fault type probability vector. For example... Figure 4 As shown, the transformer fault detection method specifically includes the following steps:

[0157] S401: Based on the dissolved gas Duval triangulation data, calculate the membership degree of each data point in the dissolved gas Duval triangulation data at each preset percentage level.

[0158] In this step, after the computer obtains the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type, in order to obtain the probability vector of the third fault type, it is necessary to calculate the membership degree of each data in the dissolved gas Duval triangulation data at each preset percentage level based on the dissolved gas Duval triangulation data.

[0159] For example, the preset percentage levels are low, medium, or high. This can be determined according to the formula. Calculate the membership degree of the ratio of methane concentration to overall gas concentration at lower levels, where, This indicates the membership degree of the ratio of methane concentration to overall gas concentration at lower levels. This represents the ratio of methane concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of methane concentration to overall gas concentration at lower levels. This represents a preset percentage range for the ratio of methane concentration to overall gas concentration at lower levels.

[0160] According to the formula Calculate the membership degree of the ratio of methane concentration to overall gas concentration at the intermediate level, where, This indicates the membership degree of the ratio of methane concentration to overall gas concentration at the intermediate level. This represents the ratio of methane concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of methane concentration to overall gas concentration at the intermediate level. This represents a preset percentage range for the ratio of methane concentration to overall gas concentration at the medium level.

[0161] According to the formula Calculate the membership degree of the ratio of methane concentration to overall gas concentration at higher levels, where, This indicates the membership degree of the ratio of methane concentration to overall gas concentration at higher levels. This represents the ratio of methane concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of methane concentration to overall gas concentration at a high level. This represents a preset percentage range for the ratio of methane concentration to overall gas concentration at higher levels.

[0162] It should be noted that It can be 0.2, 0.25, 0.3, etc. It can be 0.7, 0.75, 0.8, etc. It can be 1.2, 1.25, 1.3, etc. , and It can be 1.45, 1.5, 1.55, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0163] According to the formula Calculate the membership degree of the ratio of acetylene concentration to overall gas concentration at lower levels, where, This indicates the membership degree of the ratio of acetylene concentration to overall gas concentration at lower levels. This represents the ratio of acetylene concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of acetylene concentration to overall gas concentration at lower levels. This represents a preset percentage range for the ratio of acetylene concentration to overall gas concentration at lower levels.

[0164] According to the formula Calculate the membership degree of the ratio of acetylene concentration to overall gas concentration at the intermediate level, where, This indicates the membership degree of the ratio of acetylene concentration to overall gas concentration at the intermediate level. This represents the ratio of acetylene concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of acetylene concentration to overall gas concentration at the medium level. This represents a preset percentage range for the ratio of acetylene concentration to overall gas concentration at the medium level.

[0165] According to the formula Calculate the membership degree of the ratio of acetylene concentration to overall gas concentration at higher levels, where, This indicates the membership degree of the ratio of acetylene concentration to overall gas concentration at higher levels. This represents the ratio of acetylene concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of acetylene concentration to overall gas concentration at a high level. This represents a preset percentage range for the ratio of acetylene concentration to overall gas concentration at higher levels.

[0166] It should be noted that It can be 0.2, 0.25, 0.3, etc. It can be 0.7, 0.75, 0.8, etc. It can be 1.2, 1.25, 1.3, etc. , and It can be 1.45, 1.5, 1.55, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0167] According to the formula Calculate the membership degree of the ratio of ethylene concentration to overall gas concentration at lower levels, where, This indicates the membership degree of the ratio of ethylene concentration to overall gas concentration at lower levels. This represents the ratio of ethylene concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of ethylene concentration to overall gas concentration at lower levels. This represents a preset percentage range for the ratio of ethylene concentration to overall gas concentration at lower levels.

[0168] According to the formula Calculate the membership degree of the ratio of ethylene concentration to overall gas concentration at the intermediate level, where, This indicates the membership degree of the ratio of ethylene concentration to overall gas concentration at the intermediate level. This represents the ratio of ethylene concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of ethylene concentration to overall gas concentration at the medium level. This represents a preset percentage range for the ratio of ethylene concentration to overall gas concentration at the medium level.

[0169] According to the formula Calculate the membership degree of the ratio of ethylene concentration to overall gas concentration at higher levels, where, This indicates the membership degree of the ratio of ethylene concentration to overall gas concentration at higher levels. This represents the ratio of ethylene concentration to the overall gas concentration. This represents the preset percentage center value of the ratio of ethylene concentration to overall gas concentration at higher levels. This represents a preset percentage range for the ratio of ethylene concentration to overall gas concentration at higher levels.

[0170] It should be noted that It can be 0.7, 0.8, 0.9, etc. It could be 2.1, 2.2, 2.3, etc. It can be 3.5, 3.6, 3.7, etc. , and It can be 1.3, 1.4, 1.5, etc., but the embodiments in this application do not specify. , , , , and Limitations can be set based on the actual situation.

[0171] S402: Determine the matching degree of each fault type based on the membership degree of each data point in the dissolved gas Duval triangulation data at each preset percentage level.

[0172] In this step, after the computer obtains the membership degree of each data point in the dissolved gas Duval triangle data at each preset percentage level, it can determine the matching degree of each fault type based on these membership degrees.

[0173] Specifically, there is a corresponding relationship between each fault type and its membership degree. Therefore, for each fault type, the matching degree of the fault type can be determined based on the membership degree corresponding to that fault type.

[0174] The maximum, minimum, or average membership degree corresponding to the fault type can be used as the matching degree of the fault type.

[0175] For example, the membership degree corresponding to a partial discharge fault includes: the membership degree of the ratio of acetylene concentration to overall gas concentration at a low level, and the membership degree of the ratio of methane concentration to overall gas concentration at a high level.

[0176] The membership degrees corresponding to thermal faults include: the membership degree of the ratio of acetylene concentration to overall gas concentration at the medium level, and the membership degree of the ratio of methane concentration to overall gas concentration at the medium level.

[0177] The membership degree corresponding to the arc fault includes: the membership degree of the ratio of acetylene concentration to total gas concentration at high levels, and the membership degree of the ratio of methane concentration to total gas concentration at low levels.

[0178] S403: For each fault type, the product of the matching degree of the fault type and the third accuracy of the third fuzzy prediction model for that fault type is used as the fault probability of that fault type.

[0179] In this step, after the computer obtains the matching degree for each fault type, in order to improve the detection accuracy, for each fault type, the product of the matching degree of the fault type and the third accuracy of the third fuzzy prediction model for that fault type is used as the fault probability of that fault type.

[0180] It should be noted that the matching degree of each fault type can be normalized before calculating the fault probability of each fault type.

[0181] For example, the third fuzzy prediction model has an accuracy rate of 82.61% for partial discharge faults, 98.765% for thermal faults, and 92.309% for arc faults. This application does not limit the accuracy rate of the third fuzzy prediction model for each fault type.

[0182] S404: Generate a third fault type probability vector based on the fault probability of each fault type.

[0183] In this step, after the computer obtains the failure probability of each failure type, it combines the failure probabilities of each failure type to obtain the third failure type probability vector.

[0184] The transformer fault detection method provided in this embodiment improves the accuracy of the third fault type probability vector by using a second fuzzy prediction model to predict the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type.

[0185] Figure 5 This is a flowchart illustrating Embodiment 5 of the transformer fault detection method provided in this application. Based on the above embodiments, this application embodiment describes the accuracy of fusion detection when the detection result is the target fault type, and how the computer determines the target fault type. Figure 5 As shown, the transformer fault detection method specifically includes the following steps:

[0186] S501: If the detection result is the target fault type, then calculate the fusion detection accuracy of each fault type based on the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type.

[0187] In this step, after the computer obtains the detection results, if the detection results match the target fault type, the detection accuracy rate can be calculated to determine the detection accuracy. The fusion detection accuracy rate for each fault type can be calculated based on the first accuracy rate of the first fuzzy prediction model for each fault type, the second accuracy rate of the second fuzzy prediction model for each fault type, and the third accuracy rate of the third fuzzy prediction model for each fault type.

[0188] Specifically, for each fault type, the fusion detection accuracy of the fault type can be calculated based on the first accuracy of the first fuzzy prediction model for the fault type, the second accuracy of the second fuzzy prediction model for the fault type, and the third accuracy of the third fuzzy prediction model for the fault type.

[0189] It should be noted that the product, average, or minimum value of the first accuracy rate of the first fuzzy prediction model for the fault type, the second accuracy rate of the second fuzzy prediction model for the fault type, and the third accuracy rate of the third fuzzy prediction model for the fault type can be used as the fusion detection accuracy rate for that fault type. This application does not limit the method for calculating the fusion detection accuracy rate; it can be determined according to the actual situation.

[0190] It should be noted that the fusion detection accuracy can also be calculated only for the target fault type.

[0191] S502: Output the fusion detection accuracy of the target fault type.

[0192] In this step, after the computer obtains the fusion detection accuracy for each fault type, it outputs the fusion detection accuracy for the target fault type for the user to view.

[0193] The transformer fault detection method provided in this embodiment calculates and outputs the fusion detection accuracy of the target fault type by using the accuracy of the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model for the fault type, thereby improving the accuracy of the detection accuracy.

[0194] Figure 6 This is a flowchart illustrating Embodiment Six of the transformer fault detection method provided in this application. Based on the above embodiments, this application describes the case where a classification model is obtained through computer model training. Figure 6 As shown, the transformer fault detection method specifically includes the following steps:

[0195] S601: Select a training data point from the prediction training dataset.

[0196] In this step, in order to train a classification model, the computer needs to first predict a training data point selected from the training dataset.

[0197] Each training data set includes three training fault type probability vectors, and each training data set has a corresponding label, which is used to represent no fault or fault type.

[0198] S602: Input the training data into the initial model to obtain the training results.

[0199] In this step, after the computer selects the training data, it inputs the training data into the initial model to obtain the training results. The training results are used to characterize fault-free or fault types. The initial model is a classification model.

[0200] S603: Calculate the loss value based on the training results and the labeling of the training data.

[0201] S604: Update the initial model based on the loss value to obtain the updated model.

[0202] In the above steps, after the computer obtains the training results, it calculates a loss value based on the training results and the labels of the training data in order to update the model. Then, the initial model is updated based on the loss value to obtain the updated model.

[0203] S605: Validate the updated model using the validation dataset to determine if the validation passed. If the validation passed, proceed to step S606; if the validation failed, proceed to step S607.

[0204] In this step, after the computer obtains the updated model, in order to determine whether to continue training, it can validate the updated model against the validation dataset to determine whether the validation is successful.

[0205] For example, the validation data from the validation dataset is sequentially input into the updated model to determine the accuracy of the updated model's output. If the accuracy is greater than a preset validation threshold, the validation is considered successful; if the accuracy is less than or equal to the preset validation threshold, the validation is considered unsuccessful.

[0206] It should be noted that the preset verification threshold can be 85%, 90%, 95%, etc. This application embodiment does not limit the preset verification threshold, which can be determined according to the actual situation.

[0207] S606: Use the updated model as the classification model.

[0208] In this step, if the computer determines that the verification has passed, it means that the training is complete and no further training is needed. The updated model will then be used as the classification model.

[0209] S607: Use the updated model as the new initial model and execute step S601.

[0210] In this step, if the computer determines that the verification fails, it means that further training is needed. The updated model is then used as the new initial model, and the process returns to step S601.

[0211] In other words, new training data is selected from the prediction training dataset and input into a new initial model to obtain new training results. Based on the new training results and the new training data labels, a new loss value is calculated, and the new initial model is updated accordingly to obtain the updated model. The updated model is then validated using a validation dataset to determine if it passes validation. If validation fails, this process is repeated until validation is successful, and the updated model is then used as the classification model.

[0212] It should be noted that after each transformer fault detection using the classification model, incremental training can be performed on the classification model based on the probability vectors of the first fault type, the second fault type, and the third fault type to improve the detection accuracy of the classification model.

[0213] The transformer fault detection method provided in this embodiment improves the detection accuracy of the classification model by training an initial model with training data.

[0214] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0215] Figure 7 This is a schematic diagram of the structure of an embodiment of the transformer fault detection device provided in this application. Figure 7 As shown, the transformer fault detection device 70 includes:

[0216] Processing module 71 is used to determine the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model based on the acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangular data, model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model.

[0217] Detection module 72 is used to input the first fault type probability vector, the second fault type probability vector and the third fault type probability vector into the classification model to obtain the detection result, wherein the detection result is no fault or target fault type;

[0218] The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector all include the fault probability of each fault type.

[0219] Furthermore, the dissolved gas concentration data includes: hydrogen concentration, methane concentration, and carbon monoxide concentration;

[0220] The dissolved gas ratio data includes: the ratio of methane concentration to hydrogen concentration, and the ratio of acetylene concentration to ethylene concentration;

[0221] The dissolved gas Duval triangle data includes: the ratio of methane concentration to the overall gas concentration, the ratio of acetylene concentration to the overall gas concentration, and the ratio of ethylene concentration to the overall gas concentration; the overall gas concentration is the sum of the methane concentration, the acetylene concentration, and the ethylene concentration.

[0222] The model detection accuracy data includes: the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type.

[0223] Furthermore, the processing module 71 is specifically used for:

[0224] Using the first fuzzy prediction model, the dissolved gas concentration data, the dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type are predicted to obtain the first fault type probability vector.

[0225] Using the second fuzzy prediction model, the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type are predicted to obtain the second fault type probability vector;

[0226] Using the third fuzzy prediction model, the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type are predicted to obtain the probability vector of the third fault type.

[0227] Furthermore, the processing module 71 is specifically used for:

[0228] The following processing is performed using the first fuzzy prediction model:

[0229] Based on the dissolved gas concentration data, calculate the membership degree of each data point in the dissolved gas concentration data at each preset concentration level;

[0230] Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data under each preset ratio level;

[0231] Based on the membership degree of each data point in the dissolved gas concentration data at each preset concentration level, and the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, the matching degree of each fault type is determined;

[0232] For each fault type, the product of the matching degree of the fault type and the first accuracy of the first fuzzy prediction model for the fault type is taken as the fault probability of the fault type.

[0233] Generate the first fault type probability vector based on the fault probability of each fault type.

[0234] Furthermore, the processing module 71 is specifically used for:

[0235] The second fuzzy prediction model is used for the following processing:

[0236] Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data under each preset ratio level;

[0237] Based on the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, the matching degree of each fault type is determined;

[0238] For each fault type, the product of the matching degree of the fault type and the second accuracy of the second fuzzy prediction model for the fault type is taken as the fault probability of the fault type.

[0239] Based on the failure probability of each failure type, a second failure type probability vector is generated.

[0240] Furthermore, the processing module 71 is specifically used for:

[0241] The third fuzzy prediction model is used for the following processing:

[0242] Based on the dissolved gas Duval triangular data, calculate the membership degree of each data point in the dissolved gas Duval triangular data at each preset percentage level;

[0243] Based on the membership degree of each data point in the dissolved gas Duval triangle data at each preset percentage level, the matching degree of each fault type is determined;

[0244] For each fault type, the product of the matching degree of the fault type and the third accuracy of the third fuzzy prediction model for the fault type is taken as the fault probability of the fault type.

[0245] The third fault type probability vector is generated based on the fault probability of each fault type.

[0246] Furthermore, the processing module 71 is also used for:

[0247] If the detection result is the target fault type, then the fusion detection accuracy of each fault type is calculated based on the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type.

[0248] Output the fusion detection accuracy of the target fault type.

[0249] The transformer fault detection device provided in this embodiment is used to execute the technical solution in any of the aforementioned method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0250] Figure 8 This is a schematic diagram of the structure of an electronic device provided in this application. Figure 8 As shown, the electronic device 80 includes:

[0251] Processor 81, memory 82, and communication interface 83;

[0252] The memory 82 is used to store the executable instructions of the processor 81;

[0253] The processor 81 is configured to execute the technical solutions in any of the foregoing method embodiments by executing the executable instructions.

[0254] Optionally, the memory 82 can be either standalone or integrated with the processor 81.

[0255] Optionally, when the memory 82 is a device independent of the processor 81, the electronic device 80 may further include:

[0256] Bus 84, memory 82 and communication interface 83 are connected to processor 81 through bus 84 and complete communication with each other. Communication interface 83 is used to communicate with other devices.

[0257] Optionally, the communication interface 83 can be implemented using a transceiver. The communication interface is used to enable communication between the database access device and other devices (e.g., clients, read-write databases, and read-only databases). The memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk drive.

[0258] Bus 84 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus.

[0259] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0260] The electronic device is used to execute the technical solutions in any of the foregoing method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0261] This application also provides a readable storage medium storing a computer program thereon, which, when executed by a processor, implements the technical solutions provided in any of the foregoing method embodiments.

[0262] This application also provides a computer program product, including a computer program, which, when executed by a processor, is used to implement the technical solutions provided in any of the foregoing method embodiments.

[0263] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0264] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for detecting transformer faults, characterized in that, include: Based on the obtained dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangle data, and model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model, the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model are determined. The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector are input into the classification model to obtain the detection result, which is either no fault or the target fault type. The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector all include the fault probability of each fault type.

2. The method according to claim 1, characterized in that, The dissolved gas concentration data includes: hydrogen concentration, methane concentration, and carbon monoxide concentration; The dissolved gas ratio data includes: the ratio of methane concentration to hydrogen concentration, and the ratio of acetylene concentration to ethylene concentration; The dissolved gas Duval triangle data includes: the ratio of methane concentration to the overall gas concentration, the ratio of acetylene concentration to the overall gas concentration, and the ratio of ethylene concentration to the overall gas concentration; the overall gas concentration is the sum of the methane concentration, the acetylene concentration, and the ethylene concentration. The model detection accuracy data includes: the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type.

3. The method according to claim 2, characterized in that, The step of determining the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model based on the acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangular data, and model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model, includes: Using the first fuzzy prediction model, the dissolved gas concentration data, the dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type are predicted to obtain the first fault type probability vector. Using the second fuzzy prediction model, the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type are predicted to obtain the second fault type probability vector; Using the third fuzzy prediction model, the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type are predicted to obtain the probability vector of the third fault type.

4. The method according to claim 3, characterized in that, The first fuzzy prediction model is used to predict the dissolved gas concentration data, the dissolved gas ratio data, and the first accuracy of the first fuzzy prediction model for each fault type, to obtain the first fault type probability vector, including: The following processing is performed using the first fuzzy prediction model: Based on the dissolved gas concentration data, calculate the membership degree of each data point in the dissolved gas concentration data at each preset concentration level; Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data under each preset ratio level; Based on the membership degree of each data point in the dissolved gas concentration data at each preset concentration level, and the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, the matching degree of each fault type is determined; For each fault type, the product of the matching degree of the fault type and the first accuracy of the first fuzzy prediction model for the fault type is taken as the fault probability of the fault type. Generate the first fault type probability vector based on the fault probability of each fault type.

5. The method according to claim 3, characterized in that, The second fuzzy prediction model is used to predict the dissolved gas ratio data and the second accuracy of the second fuzzy prediction model for each fault type, resulting in a second fault type probability vector, including: The second fuzzy prediction model is used for the following processing: Based on the dissolved gas ratio data, calculate the membership degree of each data point in the dissolved gas ratio data under each preset ratio level; Based on the membership degree of each data point in the dissolved gas ratio data at each preset ratio level, the matching degree of each fault type is determined; For each fault type, the product of the matching degree of the fault type and the second accuracy of the second fuzzy prediction model for the fault type is taken as the fault probability of the fault type. Based on the failure probability of each failure type, a second failure type probability vector is generated.

6. The method according to claim 3, characterized in that, The process employs the third fuzzy prediction model to predict the dissolved gas Duval triangulation data and the third accuracy of the third fuzzy prediction model for each fault type, thereby obtaining the probability vector of the third fault type, including: The third fuzzy prediction model is used for the following processing: Based on the dissolved gas Duval triangular data, calculate the membership degree of each data point in the dissolved gas Duval triangular data at each preset percentage level; Based on the membership degree of each data point in the dissolved gas Duval triangle data at each preset percentage level, the matching degree of each fault type is determined; For each fault type, the product of the matching degree of the fault type and the third accuracy of the third fuzzy prediction model for the fault type is taken as the fault probability of the fault type. The third fault type probability vector is generated based on the fault probability of each fault type.

7. The method according to claim 2, characterized in that, The method further includes: If the detection result is the target fault type, then the fusion detection accuracy of each fault type is calculated based on the first accuracy of the first fuzzy prediction model for each fault type, the second accuracy of the second fuzzy prediction model for each fault type, and the third accuracy of the third fuzzy prediction model for each fault type. Output the fusion detection accuracy of the target fault type.

8. A transformer fault detection device, characterized in that, include: The processing module is used to determine the first fault type probability vector corresponding to the first fuzzy prediction model, the second fault type probability vector corresponding to the second fuzzy prediction model, and the third fault type probability vector corresponding to the third fuzzy prediction model based on the acquired dissolved gas concentration data, dissolved gas ratio data, dissolved gas Duval triangular data, model detection accuracy data, as well as the first fuzzy prediction model, the second fuzzy prediction model, and the third fuzzy prediction model. The detection module is used to input the first fault type probability vector, the second fault type probability vector and the third fault type probability vector into the classification model to obtain the detection result, wherein the detection result is no fault or the target fault type; The first fault type probability vector, the second fault type probability vector, and the third fault type probability vector all include the fault probability of each fault type.

9. A server, characterized in that, include: Processor, memory, communication interface; The memory is used to store the executable instructions of the processor; The processor is configured to execute the transformer fault detection method according to any one of claims 1 to 7 by executing the executable instructions.

10. A readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the transformer fault detection method according to any one of claims 1 to 7.

11. A computer program product, characterized in that, It includes a computer program, which, when executed by a processor, is used to implement the transformer fault detection method according to any one of claims 1 to 7.

Images