A method for detecting the state of a transformer test insulating piece

Abstract

This invention discloses a method for detecting the condition of transformer test insulation components. The method includes the following steps: S1, sampling the insulation paper on the surface of the transformer windings or leads, and recording and statistically analyzing the temperature rise changes in various parts of the transformer; S2, based on the temperature rise change statistics, selecting the points with the largest temperature rise changes in each winding or lead, and identifying the corresponding insulation paper; S3, conducting a cohesion degree test, obtaining data, and processing it; S4, inputting the cohesion degree test results into the insulation component detection model, outputting the detection interval, and determining the condition of the insulation component based on the detection interval. This invention solves the problem of damage to internal insulation components during existing transformer testing. It detects the condition of internal insulation components while protecting the transformer, ensuring the safety of transformer testing and facilitating continued maintenance.

Description

Technical Field

[0001] This invention relates to the field of transformer testing technology, and in particular to a method for detecting the condition of transformer insulation components. Background Technology

[0002] Transformers are one of the most important pieces of equipment in power systems. They contain a large amount of insulating paper material for insulation and isolation between internal coils and between coils and ground. Currently, transformer insulation faults account for more than 80% of all faults. Some transformer tests are destructive and can damage internal insulation components. This paper proposes a method for testing the condition of transformer insulation components, which allows for testing and inspection while protecting the transformer.

[0003] The Chinese patent document CN112834870A, entitled "A Test Device for a Valve-Side Outgoing Line Device of an Ultra-High Voltage Converter Transformer," discloses a test device for a valve-side outgoing line device of an ultra-high voltage converter transformer. It includes an oil conservator, a test riser, a test bushing, and a test oil tank. The oil conservator is positioned above the test oil tank and connected to it via a connecting pipe. The test riser is positioned on top of the test oil tank, and the test bushing is connected to it. A valve-side outgoing line device is installed inside the test riser, and a shielding device is installed inside the test oil tank. However, this Chinese patent mainly concerns the specific structure of the test device and does not address the aforementioned problem. Summary of the Invention

[0004] This invention solves the problem that existing transformer testing processes can damage internal insulation components. It proposes a method for detecting the condition of transformer insulation components, which detects the condition of internal insulation components while protecting the transformer, ensuring the safety of transformer testing and facilitating continued maintenance.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for detecting the condition of transformer test insulation components, comprising the following steps:

[0006] S1, sample the insulating paper on the surface of the transformer windings or leads, and record and statistically analyze the temperature rise changes in various parts of the transformer;

[0007] S2. Based on the temperature rise statistics, select the point with the largest temperature rise change for each winding or lead, and screen out the corresponding insulating paper.

[0008] S3, Conduct polymerization degree experiments, obtain data and process it;

[0009] S4 inputs the polymerization degree test results into the insulation component detection model, outputs the detection range, and determines the state of the insulation component based on the detection range.

[0010] In this invention, the insulating paper on the surface of the transformer windings or leads is sampled, and then the temperature rise changes inside the transformer are statistically analyzed. Based on the statistical results, several points with the largest temperature rise changes and the corresponding insulating paper are selected. Then, a polymerization degree test is performed. After removing relevant data, the average value is calculated to obtain the polymerization degree test result. Finally, the polymerization degree test result is put into the insulation component detection model and the final detection range is output to determine the state of the insulation component. The insulation component is preset with multiple states, namely normal state, slightly damaged state, damaged state, and severely damaged state.

[0011] Preferably, step S1 includes the following steps:

[0012] S11, peel off the insulating paper and secondary insulating paper on the surface of the transformer winding or lead wire to take samples, and divide each insulating paper sample and secondary insulating paper sample into several equal parts.

[0013] S12 records and statistically analyzes the temperature rise changes of various parts of the transformer under rated current, and stores the statistical results.

[0014] In this invention, the recording and statistics of temperature rise changes can be performed on a computer or other mobile device, and after the statistics are completed, they can be stored on a computer or other mobile device and retrieved at any time.

[0015] Preferably, step S2 specifically includes:

[0016] Based on the temperature rise statistics, a temperature rise change chart is drawn with the temperature rise change value and each part as the main characteristics. Several points with the largest temperature rise change in each winding or lead are selected, along with the corresponding insulating paper.

[0017] In this invention, after drawing the statistical chart of temperature rise changes, a separate chart can also be drawn for a specific part of the transformer.

[0018] Preferably, step S3 specifically involves: obtaining preliminary polymerization degree test data through polymerization degree tests, eliminating failed polymerization degree tests and abnormal data in polymerization degree tests, and calculating the average value of the remaining polymerization degree test data for each part.

[0019] In this invention, the degree of polymerization test is an existing technology that can obtain preliminary test data on the degree of polymerization. After a series of elimination processes on the data, the average value of the remaining data is calculated, thus completing the data processing process.

[0020] Preferably, step S4 includes the following steps:

[0021] S41. Establish an insulation component detection model. This model is based on LSTM, with the aggregation degree test results as input and the detection range as output. The detection range is specifically the range of transformer damage degree.

[0022] S42 determines the final state of the insulation component by matching the detection range with the state of the insulation component, in order to decide whether to replace or modify it.

[0023] In this invention, the insulation component detection model is built using LSTM, and the model is trained using historical aggregation degree test results and historical detection intervals as training sets. The model has deep learning capabilities to ensure the accuracy of data detection. The detection intervals correspond one-to-one with the insulation component states, and the correspondence between the two is determined by superimposing and statistically analyzing historical experience data.

[0024] Preferably, in step S41, the preliminary result output by the LSTM model is the detection value, and the average value of the detection value is... As an estimate of the true regression value, its variance σ is also calculated. 2 (X), and calculate the final detection interval, specifically:

[0025]

[0026] Where α is the confidence level, z α This is the critical value corresponding to a confidence level of α in the standard Gaussian function.

[0027] In this invention, compared to individual detection values, the output detection range has reference value. The detection range includes the fluctuation range of the calculated detection values, which can better correspond to the state of the insulating component. If the detection range falls within the intersection range of two insulating component states, the state with more severe damage between the two should be used as the standard.

[0028] The beneficial effects of the present invention are as follows: the transformer test insulation component condition detection method of the present invention, while protecting the transformer, can detect the condition of the internal insulation components, promptly detect internal problems of the transformer, ensure the safety of transformer testing, facilitate maintenance and formulate corresponding response strategies. Attached Figure Description

[0029] Figure 1 This is a flowchart of a method for detecting the condition of transformer test insulation components according to the present invention. Detailed Implementation

[0030] Example:

[0031] This embodiment proposes a method for detecting the condition of transformer test insulation components, referring to... Figure 1 This includes the following steps.

[0032] Step S1 involves sampling the insulating paper on the surface of the transformer windings or leads, and recording and statistically analyzing the temperature rise changes in various parts of the transformer. The transformer contains a large amount of insulating paper. During transformer operation, the solid insulating material slowly degrades through various pathways such as hydrolysis, pyrolysis, and oxidation, causing the internal highly polymerized long-chain structure to break down. This manifests as a decrease in the degree of polymerization of the insulating paper and the appearance of furfural and dissolved gases in the oil. While the furfural and dissolved gas detection can be performed under energized conditions, it is affected by other fault factors and cannot accurately determine the degree of aging of the insulating paper. Therefore, the degree of polymerization is closely related to the electrical characteristics of the insulating paper. Specifically, this step includes the following two sub-steps.

[0033] First, proceed to step S11, peeling off the insulating paper and secondary insulating paper from the surface of the transformer windings or leads to take samples. Divide each insulating paper sample and secondary insulating paper sample into several equal parts. Specifically, for sampling, a comprehensive sampling method is adopted. Although this sampling method increases the workload, it can ensure comprehensive sampling of the transformer windings or leads. Alternatively, the point with the largest temperature rise change can be determined first, and then the insulating paper and secondary insulating paper can be sampled. However, this step cannot eliminate the influence of statistical errors in temperature rise change.

[0034] Then, in step S12, the temperature rise changes of various parts of the transformer under rated current are recorded, statistically analyzed, and the results are stored. Specifically, this step is performed on a computer or other mobile device.

[0035] Step S2: Based on the temperature rise change statistics, select the points with the largest temperature rise changes for each winding or lead, and then select the corresponding insulating paper. Specifically, in this step, based on the temperature rise change statistics, using the temperature rise change value and each part as the main features, draw a temperature rise change statistical chart, and select several points with the largest temperature rise changes for each winding or lead, along with the corresponding insulating paper. In this embodiment, 10 points with the largest temperature rise changes are selected for each winding or lead.

[0036] Step S3: Conduct polymerization degree experiments, obtain data, and process it. Specifically, in this step, preliminary polymerization degree experiment data is obtained through polymerization degree experiments. Failed polymerization degree experiments and abnormal data in the polymerization degree experiments are eliminated, and the average value of the remaining polymerization degree experiment data for each part is calculated. Specifically, failed polymerization degree experiments and their corresponding data are completely eliminated. Some abnormal data in the experiment, such as data that are obviously inconsistent with common sense or have excessively large mutation values, should also be eliminated.

[0037] Step S4: Input the polymerization degree test results into the insulation component detection model, output the detection range, and determine the state of the insulation component based on the detection range; specifically, this step includes the following two sub-steps.

[0038] Step S41: Establish an insulation component detection model. This model is based on LSTM, with the aggregation degree test results as input and the detection range as output. The detection range is specifically the range of transformer damage. In particular, the detection range is a range value, which greatly reduces the problem of inaccurate detection.

[0039] Step S42: By correlating the detection interval with the state of the insulating component, the final state of the insulating component is determined to decide whether to replace or modify it; specifically, the correspondence between the detection interval and the state of the insulating component is determined by superimposing and statistically analyzing historical experience data.

[0040] In step S41, the preliminary results output by the LSTM model are the detection values, and the average value of the detection values ​​is... As an estimate of the true regression value, its variance σ is also calculated. 2 (X), and calculate the final detection interval, specifically:

[0041]

[0042] Where α is the confidence level, z α This represents the critical value corresponding to a confidence level of α in the standard Gaussian function. Specifically, for the average value of the detected values... It is related to the detected value and can be calculated using the following formula:

[0043]

[0044] Among them, h k (X) represents the preliminary result output by the model, which is the detection value;

[0045] And for variance σ 2 The calculation of (X) is given by the following formula:

[0046]

[0047] Therefore, the detection interval can be calculated.

[0048] In this embodiment, the insulating paper on the surface of the transformer windings or leads is sampled, and then the temperature rise changes inside the transformer are statistically analyzed. Based on the statistical results, several points with the largest temperature rise changes and the corresponding insulating paper are selected. Then, a polymerization degree test is performed. After removing relevant data, the average value is calculated to obtain the polymerization degree test result. Finally, the polymerization degree test result is put into the insulation component detection model and the final detection range is output to determine the state of the insulation component. The insulation component is preset with multiple states, namely normal state, slightly damaged state, damaged state, and severely damaged state.

[0049] In this embodiment, the recording and statistics of temperature rise changes can be performed on a computer or other mobile device, and after the statistics are completed, they can be stored on a computer or other mobile device for retrieval at any time.

[0050] In this embodiment, after drawing the statistical chart of temperature rise changes, a separate chart can also be drawn for a specific part of the transformer so that a specific part can be analyzed.

[0051] In this embodiment, the degree of polymerization test is existing technology and will not be described in detail here. In this embodiment, the degree of polymerization test can obtain preliminary test data of the degree of polymerization. After a series of elimination processes on the data, the average value of the remaining data is calculated, thus completing the data processing process.

[0052] In this embodiment, the insulation component detection model is built using LSTM. Specifically, the model is trained using historical aggregation degree test results and historical detection intervals as the training set. After training, data can be input into the model for processing. The model has deep learning capabilities to ensure the accuracy of data detection. The detection interval corresponds one-to-one with the insulation component state, and the correspondence between the two is determined by superimposing and statistically analyzing historical experience data.

[0053] In this embodiment, compared to a single detection value, the output detection range can be of reference value. The detection range includes the fluctuation range of the calculated detection value, which can better correspond to the state of the insulating component. If the detection range falls within the intersection range of two insulating component states, the state with more severe damage between the two should be taken as the standard.

[0054] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method of detecting the state of an insulation piece for a transformer test, characterized by, Includes the following steps: S1, sample the insulating paper on the surface of the transformer windings or leads, and record and statistically analyze the temperature rise changes in various parts of the transformer; S2. Based on the temperature rise statistics, select the point with the largest temperature rise change for each winding or lead, and screen out the corresponding insulating paper. S3, Conduct polymerization degree experiments, obtain data and process it; S4 inputs the polymerization degree test results into the insulation component inspection model, outputs the inspection interval, and determines the state of the insulation component based on the inspection interval; S4 includes: An insulation piece detection model based on LSTM is established, and the preliminary result output by the model is a detection value As an estimated true regression value, its variance is calculated , the detection interval is calculated Subtract to Add The interval formed by is the critical value corresponding to the confidence level in the standard Gaussian function.

2. The method for detecting the condition of transformer test insulation components according to claim 1, characterized in that, Step S1 includes the following steps: S11, peel off the insulating paper and secondary insulating paper on the surface of the transformer winding or lead wire to take samples, and divide each insulating paper sample and secondary insulating paper sample into several equal parts. S12 records and statistically analyzes the temperature rise changes of various parts of the transformer under rated current, and stores the statistical results.

3. A method for detecting the condition of transformer test insulation components according to claim 1 or 2, characterized in that, Step S2 specifically involves: Based on the temperature rise change statistics, a temperature rise change statistical chart is drawn with the temperature rise change value and each part as the main characteristics. Several points with the largest temperature rise change in each winding or lead are selected, as well as the insulation paper corresponding to the positions of the several points with the largest temperature rise change in each winding or lead.

4. The method for detecting the condition of transformer test insulation components according to claim 3, characterized in that, Step S3 specifically involves: obtaining preliminary polymerization degree test data through polymerization degree tests, eliminating failed polymerization degree tests and abnormal data in polymerization degree tests, and calculating the average value of each remaining polymerization degree test data.

5. A method for detecting the condition of transformer test insulation components according to claim 1 or 2, characterized in that, Step S4 includes the following steps: S41. Establish an insulation component detection model. This model is based on LSTM, with the aggregation degree test results as input and the detection range as output. The detection range is specifically the range of transformer damage degree. S42 determines the final state of the insulation component by matching the detection range with the state of the insulation component, in order to decide whether to replace or modify it.

6. The method for detecting the condition of transformer test insulation components according to claim 5, characterized in that, In step S41, the insulation component detection model is trained using historical aggregation degree test results and historical detection intervals as the training set, and the model has deep learning capabilities.

7. The method for detecting the condition of transformer test insulation components according to claim 3, characterized in that, After drawing the statistical chart of temperature rise changes, draw a separate chart for a specific part of the transformer.

Images