Transformer typical defect simulation and diagnosis system and method based on fusion algorithm

Abstract

The invention discloses a transformer typical defect simulation and diagnosis system based on a fusion algorithm, and the system carries out the iterative training of multi-parameter feature data corresponding to various transformer typical defects under different environment parameters through the fusion algorithm, and generates a transformer defect fault feature database. The Euclidean distance between the real-time multi-parameter characteristic data of the to-be-diagnosed transformer and the multi-parameter characteristic data corresponding to the typical defects of various transformers under the same environmental parameters in the transformer defect fault characteristic database is calculated, and the Euclidean distance is utilized to perform similarity analysis to obtain the defect type of the to-be-diagnosed transformer. According to the invention, the accuracy and reliability of transformer typical defect detection are improved.

Description

Technical Field

[0001] This invention relates to the field of transformer defect diagnosis technology, specifically to a transformer typical defect simulation and diagnosis system and method based on a fusion algorithm. Background Technology

[0002] Transformers, as key equipment in power systems, undertake the important tasks of voltage conversion and power transmission. The operating status of transformers plays a crucial role in ensuring the safe and stable operation of the power grid and the reliable supply of electricity. Once an insulation fault occurs in a transformer, it will have a serious impact on the safe operation of the power grid, potentially triggering major power accidents, disrupting normal production, and reducing economic benefits.

[0003] However, with the gradual expansion of power system scale and the increasing complexity of its operation, the operating conditions faced by transformers are becoming increasingly complex and diverse. Simultaneously, the types of defects are also showing a continuous upward trend. Traditional transformer defect diagnosis technologies have limitations, making it difficult to integrate defect simulation, condition monitoring, and defect diagnosis. Furthermore, most defect detection technologies neglect the differences in natural conditions under which transformers operate, resulting in insufficient accuracy in defect detection. Summary of the Invention

[0004] The purpose of this invention is to provide a system and method for simulating and diagnosing typical transformer defects based on a fusion algorithm. This invention improves the accuracy and reliability of detecting typical transformer defects.

[0005] To achieve this objective, the present invention provides a transformer typical defect simulation and diagnosis system based on a fusion algorithm, comprising: The Euclidean distance calculation module is used to calculate the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault characteristic database. The defect diagnosis module is used to perform similarity analysis using Euclidean distance to determine the defect type of the transformer to be diagnosed.

[0006] Preferably, the transformer defect fault feature database is obtained by iteratively training multi-parameter feature data corresponding to various typical transformer defects under different environmental parameters using a fusion algorithm.

[0007] Preferably, the various typical defects of transformers include transformer core loosening, transformer inter-turn short circuit, transformer core surface burr discharge, and transformer clamp bolt loosening discharge; the multi-parameter characteristic data refers to the set of measured data corresponding to multiple characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition.

[0008] Preferably, the fusion algorithm includes the K-Means clustering algorithm; The fusion algorithm includes the K-Means clustering algorithm; The different environmental parameters include humidity and temperature. The multi-parameter feature data corresponding to various typical defects of transformers under different environmental parameters are classified according to the temperature and humidity levels using the K-Means clustering algorithm, resulting in four cluster databases: high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster. In the four cluster databases of high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster, a data point is randomly selected as the initial cluster center for each cluster. The data point includes temperature and humidity. Based on the objective function of the K-Means clustering algorithm, the Euclidean distance from each data point to the initial cluster center of each cluster is calculated, and the data point is assigned to the cluster database corresponding to the initial cluster center with the smallest Euclidean distance. Calculate the average value of the data points for the high temperature and high humidity cluster, the high temperature and low humidity cluster, the low temperature and high humidity cluster, and the low temperature and low humidity cluster respectively, and use them as the new cluster centers; Based on the objective function of the K-Means clustering algorithm, the Euclidean distance from each data point to the new cluster center of each cluster is calculated using the new cluster centers. The data point is then assigned to the cluster database corresponding to the new cluster center with the smallest Euclidean distance. The average value of the data points in the high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster is recalculated and used as the cluster center for each data point in the next iteration. This process continues until the cluster centers no longer change. At this point, each data point is assigned to either the high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, or low temperature and low humidity cluster, thus completing the partitioning of the cluster database.

[0009] Preferably, the fusion algorithm further includes the TOPSIS multi-criteria decision algorithm; Each cluster database contains a data sample containing multi-parameter feature data corresponding to a typical defect of a transformer. Each cluster database contains a data sample containing multi-parameter feature data corresponding to a typical defect of a transformer. Combining the TOPSIS multi-criteria decision algorithm, based on the detection results of all feature parameters in each cluster database, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition, correction factors for each feature parameter in the high-temperature high-humidity cluster, high-temperature low-humidity cluster, low-temperature high-humidity cluster, and low-temperature low-humidity cluster are calculated. The mean and standard deviation of vibration signal, port current, voltage and power, pulse current, high-frequency partial discharge and oil chromatographic components in each cluster database are calculated respectively. By calculating the mean and standard deviation of each characteristic parameter in the specific cluster database, the coefficient of variation of each characteristic parameter in the specific cluster database is obtained. Based on the coefficient of variation of each characteristic parameter in the specific cluster database, the correction factor of each characteristic parameter is calculated. Weights are assigned to each characteristic parameter index in the high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster using the entropy weight method. Organize all data samples from a specific cluster database into a data matrix. X Specific cluster databases contain n There are 10 data samples, each containing all feature parameters; Among them, matrix X This represents the multi-parameter characteristic data corresponding to all typical transformer defects collected under specific environmental conditions; x nm Indicates the first n The first data sample m The numerical values ​​of the characteristic parameters; The values ​​of each feature parameter are normalized and scaled to the [0, 1] interval to calculate the entropy value of each feature parameter, and the difference coefficient of each feature parameter is calculated based on the entropy value of each feature parameter. The difference coefficients of each characteristic parameter are normalized, and the index weights of port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal and oil chromatographic components in each cluster database are calculated respectively. The index weights of port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic components in each cluster database are multiplied by the corresponding correction factors to obtain the fusion weights of port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic components in each cluster database. The feature data in each feature parameter are weighted according to the fusion weight of each feature parameter. The feature data in each feature parameter is multiplied by the corresponding fusion weight of the feature parameter to obtain a weighted dataset. The weighted dataset, environmental parameters, and defect labels are used to train an evaluation model through a regression algorithm to learn the mapping relationship between the feature data in each feature parameter and the defect type. The evaluation model is used to predict the predicted value for each data sample. The error between the predicted value and the actual value of the defect label is calculated. If the error is greater than A, the data sample is marked as erroneous data and removed. After removing erroneous data, a new dataset is obtained. The fusion weight of each feature parameter is recalculated, and the above process is repeated. The training is iterated until the maximum number of iterations is reached, and finally, a transformer defect fault feature database is established.

[0010] Preferably, the Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database is calculated. Similarity analysis using the Euclidean distance is then performed to obtain the defect type of the transformer to be diagnosed. Specifically, this is used for: Calculate the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect and fault characteristic database. d(x,y) The calculation formula is: in, x j This represents the first of the real-time multi-parameter characteristic data of the transformer to be diagnosed. j Feature data in the feature parameters; y j This represents the first of the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database. j Feature data in the feature parameters; m It represents seven characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition. European distance d(x,y) The smaller the value, the higher the similarity. The Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database is compared, and the Euclidean distance is selected. d(x,y) The smallest typical transformer defect is used as the defect type of the transformer to be diagnosed.

[0011] This invention also provides a method for simulating and diagnosing typical transformer defects based on a fusion algorithm, comprising: The Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect and fault characteristic database is calculated. The type of defect in the transformer to be diagnosed is obtained by similarity analysis using Euclidean distance.

[0012] The present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of a transformer typical defect simulation and diagnosis method based on a fusion algorithm as described above.

[0013] The present invention also provides an electronic device, including a memory and a processor: the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, wherein when the computer-executable instructions are executed by the processor, the steps of a transformer typical defect simulation and diagnosis method based on a fusion algorithm as described above are implemented.

[0014] The beneficial effects of this invention are: This invention effectively handles the complexity of environmental parameters and multi-parameter feature data through a fusion algorithm, achieving a high degree of integration of defect simulation, defect detection, multi-parameter feature data acquisition and processing, transformer defect fault feature database training, and defect diagnosis. This results in a comprehensive and accurate simulation and diagnosis of typical transformer defects, improving the reliability and efficiency of defect detection, providing important technical support for the safe operation of transformers, and reducing economic costs and time consumption.

[0015] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2 This is a flowchart of the present invention; Figure 3 Diagram of a simulation test setup for typical defects in transformer core clamping components; Figure 4 This is a simulation diagram of burr discharge defects on the surface of the iron core; Figure 5Circuit diagram for simulation test of transformer core clamps and magnetic circuit defects; Figure 6 shows the test results of burr discharge defect detection on the iron core surface, where Figure 6(a) is the vibration detection result; Figure 6(b) is the high-frequency partial discharge detection result; Figure 6(c) is the pulse current and ultrasonic partial discharge detection result; Figure 6(d) is the port voltage, current and power detection result; and Figure 6(e) is the oil chromatography analysis result. Detailed Implementation

[0017] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0018] Example 1 A transformer typical defect simulation and diagnosis system based on fusion algorithm, such as Figure 1 As shown, it includes: The Euclidean distance calculation module is used to calculate the Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect and fault feature database. This design considers the comprehensive influence of all feature parameters, avoids misjudgment that may be caused by relying on a single feature parameter, and ensures the comprehensiveness and consistency of the data. By calculating the Euclidean distance, the difference between the multi-parameter feature data is transformed into a quantifiable distance value, which can measure the similarity between the real-time multi-parameter feature data of the transformer to be diagnosed and the data in the transformer defect and fault feature database. The defect diagnosis module is used to perform similarity analysis using Euclidean distance to obtain the defect type of the transformer to be diagnosed. This design identifies the defect type by performing similarity analysis using Euclidean distance, which can quickly match the most likely defect type, realize fast and accurate defect diagnosis, and improve the efficiency of defect detection.

[0019] In the above technical solution, the transformer defect fault feature database (which includes environmental parameters, multi-parameter feature data, and corresponding defect labels) is obtained by iteratively training the multi-parameter feature data corresponding to various typical transformer defects under different environmental parameters (different environmental parameters include humidity and temperature, which are achieved by adjusting the environment of the test platform, with a temperature range of 10-70℃ and a humidity range of 10%-70%RH) through a fusion algorithm. The above design collects multi-parameter feature data corresponding to various typical transformer defects under different environmental parameters and performs iterative training, which can ensure that the transformer defect fault feature database comprehensively covers a variety of defect conditions, providing a solid data foundation for defect diagnosis. In the above technical solution, the various typical transformer defects include transformer core loosening, transformer inter-turn short circuit, transformer core surface burr discharge, and transformer clamp bolt loosening discharge. The multi-parameter characteristic data refers to the set of measured data corresponding to multiple characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition. The above design comprehensively covers multiple typical transformer defect types, realistically reproducing the defect operating state under different environmental parameters (such as temperature and humidity changes), which can improve the accuracy, reliability, and applicability of defect diagnosis. Through multi-parameter characteristic data, comprehensive and multi-dimensional defect state information can be provided, truly reflecting the operating characteristics of the transformer under different environmental parameters (such as temperature and humidity changes), overcoming the limitations of single characteristic parameter diagnosis. Among them, port voltage and port current can characterize electrical performance, vibration signal can detect mechanical loosening, and oil chromatographic composition can reveal insulation degradation. Through the synergy of multiple characteristic parameters, subtle differences in defects can be captured, avoiding misjudgment and improving the accuracy and reliability of defect diagnosis. The transformer core loosening can be simulated by adjusting the nuts on both sides of the structural component to apply different preloads to the core. The transformer can be simulated by artificially rounding the corners of two high-voltage windings to destroy the integrity of the turn insulation, causing the turn insulation to break (exposing the copper conductor). One end of a copper wire is connected to and fixed at one turn insulation breakage location, while the other end of the copper wire is loosely connected or not in contact with another turn insulation breakage location. The copper wire is wrapped with insulating material and the tip is exposed. The burr discharge on the surface of the transformer core can be simulated by fixing a thin copper wire in the high field strength region of the core directly opposite the high voltage winding. The transformer experienced a discharge due to loose clamp bolts. This can be simulated by retaining the original transformer core and clamp grounding bushing, adding a clamp grounding bushing hole on the transformer tank cover, installing the grounding bushing, connecting one end of a copper wire to the high-voltage end, and connecting the other end to a bolt that is suspended and fixed to the outside of the clamp.

[0020] In the above technical solution, the fusion algorithm includes the K-Means clustering algorithm; The different environmental parameters include humidity and temperature. The multi-parameter feature data corresponding to various typical defects of transformers under different environmental parameters are classified according to the temperature and humidity levels using the K-Means clustering algorithm, resulting in four cluster databases: high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster. In each of the four cluster databases—high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster—one data point is randomly selected as the initial cluster center for each cluster. The data point includes temperature and humidity (the initial cluster center for the high temperature and high humidity cluster can be set to (50℃, 60%RH), the initial cluster center for the high temperature and low humidity cluster can be set to (50℃, 40%RH), the initial cluster center for the low temperature and high humidity cluster can be set to (20℃, 60%RH), and the initial cluster center for the low temperature and low humidity cluster can be set to (20℃, 40%RH)). The objective function of the Means clustering algorithm is... J for: in, J Indicates the first i The sum of squared errors within each cluster, 1 ≤ i ≤ K , i Take the integer. J The smaller the value, the more... i The better the clustering effect of data points within a cluster; K Indicates the number of clusters, K =4; C i Indicates belonging to the first i The set of all data points in a cluster; x For the first i Data points within a cluster; μ i For the first i The cluster center of each cluster; Indicates the first i Data points of each cluster x With cluster center μ i The square of the Euclidean distance between them; According to the objective function J Calculate the Euclidean distance from each data point to the initial cluster center of each cluster, and assign the data point to the cluster database corresponding to the initial cluster center with the smallest Euclidean distance; Calculate the average value of the data points for the high temperature and high humidity cluster, the high temperature and low humidity cluster, the low temperature and high humidity cluster, and the low temperature and low humidity cluster respectively, and use them as the new cluster centers; According to the objective function JThe Euclidean distance from each data point to the new cluster center is calculated using the new cluster center. The data point is then assigned to the cluster database corresponding to the new cluster center with the smallest Euclidean distance. The average value of the data points in the high-temperature and high-humidity cluster, high-temperature and low-humidity cluster, low-temperature and high-humidity cluster, and low-temperature and low-humidity cluster is recalculated and used as the cluster center for each data point in the next iteration. This process continues until the cluster centers no longer change. At this point, each data point is assigned to either the high-temperature and high-humidity cluster, high-temperature and low-humidity cluster, low-temperature and high-humidity cluster, or low-temperature and low-humidity cluster, thus completing the partitioning of the cluster database. The above design considers the comprehensive influence of all feature parameters, avoids misjudgments that may be caused by relying on a single feature parameter, and ensures the comprehensiveness and consistency of the data.

[0021] In the above technical solution, the fusion algorithm also includes the TOPSIS multi-criteria decision algorithm; Each cluster database contains a data sample containing multi-parameter feature data corresponding to a typical defect of a transformer. Combining the TOPSIS multi-criteria decision algorithm, based on the detection results of all feature parameters in each cluster database, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition, correction factors for each feature parameter in the high-temperature high-humidity cluster, high-temperature low-humidity cluster, low-temperature high-humidity cluster, and low-temperature low-humidity cluster are calculated. Calculate the mean and standard deviation of vibration signal, port current, voltage and power, pulse current, high-frequency partial discharge, and oil chromatographic components for each cluster database. The formula for calculating the mean of each characteristic parameter in a specific cluster database is as follows: in, μ j Represents the first in a specific cluster database j The mean of the characteristic parameters, 1≤ j ≤7, j Take the integer part; n This indicates the total number of data samples contained in the cluster database; x jk Indicates the first k In the data sample, the first j The numerical values ​​of the characteristic parameters; The formula for calculating the standard deviation of each feature parameter in a specific cluster database is as follows: in, σ j Represents the first in a specific cluster database j The standard deviation of the characteristic parameters; By calculating the mean and standard deviation of each feature parameter in a specific cluster database, the coefficient of variation for each feature parameter in that specific cluster database is obtained. CV j The calculation formula is: coefficient of variation CV j The larger the value, the greater the relative volatility of this feature parameter in a specific cluster database, and the lower the reliability. Based on the coefficient of variation of each feature parameter in a specific cluster database, the correction factor for each feature parameter is calculated. c j Correction factor c j The calculation formula is: Finally, the correction factor for each characteristic parameter in the high temperature and high humidity cluster, the high temperature and low humidity cluster, the low temperature and high humidity cluster, and the low temperature and low humidity cluster is obtained; Weights are assigned to each characteristic parameter index in the high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster using the entropy weight method. Organize all data samples from a specific cluster database into a data matrix. X Specific cluster databases contain n There are 10 data samples, each containing all feature parameters; Among them, matrix X This represents the multi-parameter characteristic data corresponding to all typical transformer defects collected under specific environmental conditions; x nm Indicates the first n The first data sample m The numerical values ​​of the characteristic parameters; The values ​​of each feature parameter are normalized and scaled to the [0, 1] interval. The normalization formula is as follows: Where, min( x j ) indicates the first j The minimum value of a feature parameter in all data samples, max( x j ) indicates the first j The maximum value of a feature parameter across all data samples. p ij Indicates the first i The first data sample j The normalized values ​​of the feature parameters; Calculate the entropy value of each feature parameter based on the normalized value. e j The calculation formula is: in, s It is a constant. ; n Indicates the total number of data samples; e j The entropy value represents the value of each feature parameter, reflecting the first... j The degree of dispersion of the feature data distribution among the feature parameters; Based on the entropy value of each feature parameter e j Calculate the difference coefficient for each characteristic parameter. d j : The difference coefficient of each feature parameter d j Perform normalization and calculate the index weight for each feature parameter. w j : in, m Represents 7 characteristic parameters, d j This represents the difference coefficient for each feature parameter; This allows us to obtain the index weights for port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic components in each cluster database. w j ; Weight the indicators of port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic components in each cluster database. w j With the corresponding correction factor c j Multiplying these values ​​yields the fusion weights for port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic components in each cluster database. The formula for calculating the fusion weights is as follows: in, This represents the fusion weight for each feature parameter. c j This represents the correction factor for each feature parameter. w j The index weight represents the weight of each feature parameter; The feature data for each feature parameter are weighted according to the fusion weight of each feature parameter, and the feature data for each feature parameter is multiplied by the corresponding fusion weight to obtain a weighted dataset. Using the weighted dataset, environmental parameters, and defect labels, an evaluation model is trained using a regression algorithm to learn the mapping relationship between the feature data for each feature parameter and the defect type. The evaluation model is then used to predict the value for each data sample. Calculate the predicted value Compared with the actual value of the defect label y i Error between E i : If error E i If the value of A is greater than A (A = 0.1), the data sample is marked as erroneous and removed. After removing erroneous data, a new dataset is obtained. The fusion weight of each feature parameter is recalculated, and the above process is repeated. The training is iterated until the maximum number of iterations is reached (the maximum number of iterations is usually set between 50 and 200, and can be set to 100). Finally, a transformer defect fault feature database containing different environmental parameters and defect types is established. The above design, by combining the divided cluster database with the TOPSIS multi-criteria decision algorithm to assign fusion weights to each feature parameter, and by iteratively removing obvious erroneous data (such as data samples with errors exceeding the threshold), can reduce the impact of experimental irregularities and data noise, enabling the transformer defect fault feature database to comprehensively cover defect features under different working conditions, thereby improving the reliability and adaptability of the transformer defect fault feature database.

[0022] In the above technical solution, the Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database is calculated. The Euclidean distance is then used for similarity analysis to obtain the defect type of the transformer to be diagnosed. Specifically, this is used for: Calculate the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect and fault characteristic database. d(x,y) The calculation formula is: in, x j This represents the first of the real-time multi-parameter characteristic data of the transformer to be diagnosed. j Feature data in the feature parameters; y jThis represents the first of the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database. j Feature data in the feature parameters; m It represents seven characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition. European distance d(x,y) The smaller the value, the higher the similarity. The Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database is compared, and the Euclidean distance is selected. d(x,y) The smallest typical transformer defect is used as the defect type of the transformer to be diagnosed. The above design can quantify the similarity between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters recorded in the transformer defect and fault feature database through Euclidean distance calculation. Selecting the defect type with the smallest distance as the defect type of the transformer to be diagnosed can avoid the limitations of relying on a single feature parameter or subjective judgment. Through joint analysis of multi-parameter feature data (such as seven feature parameters), the robustness of defect diagnosis is enhanced. Under complex environmental conditions, it can also effectively distinguish similar defects, improve the accuracy and reliability of defect diagnosis results, and achieve rapid and accurate defect diagnosis.

[0023] Example 2 like Figure 2 As shown, this embodiment provides a method for simulating and diagnosing typical transformer defects based on a fusion algorithm, including the following steps: S1. By calculating the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault characteristic database, specifically including: Calculate the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect and fault characteristic database. d(x,y) The calculation formula is: in, x j This represents the first of the real-time multi-parameter characteristic data of the transformer to be diagnosed. j Feature data in the feature parameters; y j This represents the first of the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database. j Feature data in the feature parameters; mIt represents seven characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition.

[0024] S2. Similarity analysis using Euclidean distance is used to determine the defect types of the transformer to be diagnosed, specifically including: European distance d(x,y) The smaller the value, the higher the similarity. The Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database is compared, and the Euclidean distance is selected. d(x,y) The smallest typical transformer defect is used as the defect type of the transformer to be diagnosed.

[0025] Example 3 Taking the test of burr discharge defects on the surface of transformer core as an example: Step 1: Connect and assemble the test power control module, defect simulation module, data measurement module, data processing module, and defect diagnosis module to establish a typical defect simulation and diagnosis system, such as... Figure 3 As shown; The test power supply control module includes a partial discharge isolation filter, a three-phase frequency converter, and a partial discharge-free step-up transformer. The defect simulation module includes a core loosening defect simulation module, an inter-turn short circuit defect simulation module, a core surface burr discharge defect simulation module, and a clamp bolt loosening discharge defect simulation module. The data measurement module includes a vibration detection module, a port current, voltage and power detection module, a pulse current and ultrasonic partial discharge detection module, a high frequency partial discharge detection module and an oil chromatography online monitoring module; The data processing module is used to iteratively train the multi-parameter feature data corresponding to various typical transformer defects under different environmental parameters through a fusion algorithm, and generate a transformer defect fault feature database. The defect diagnosis module is used to receive real-time multi-parameter feature data of the transformer to be diagnosed, and to identify the defect type of the transformer to be diagnosed through similarity analysis.

[0026] Step 2: Using a defect simulation module, fix a thin copper wire in the high field strength region of the iron core towards the high-voltage winding to simulate burr discharge defects on the iron core surface, such as... Figure 4 As shown; based on the transformer core clamping and magnetic circuit defect simulation test circuit, a transformer no-load test was conducted under the operating condition of burr discharge defects on the core surface, as follows. Figure 5 As shown.

[0027] Step 3: After all data stabilizes, based on the data measurement module, use an accelerometer, power analyzer, partial discharge analyzer, and oil chromatogram analyzer to measure and record the test port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatogram composition of each phase under different test voltages. The detection results are shown in Figures 6(a), 6(b), 6(c), 6(d), and 6(e), and the temperature and humidity are also recorded. Port voltage, port current, and reactive power loss are measured by the port current, voltage, and power detection module; pulse current is measured by the pulse current and ultrasonic partial discharge detection module; high-frequency partial discharge is measured by the high-frequency partial discharge detection module; vibration signal is measured by the vibration detection module; and oil chromatographic components are measured by the oil chromatographic online monitoring module. The vibration detection module includes five accelerometers, a data acquisition card, and a host computer. The accelerometers are piezoelectric sensors that are magnetically fixed to the outer casing and oil tank cover of the transformer. The measurement range is 0.2Hz to 4000Hz, the measurement sensitivity is 1 V / (10 m / s2), and the range is ±5g. The data acquisition card has a sampling frequency of 128kHz and can realize the synchronous acquisition of 8 signals. The port current, voltage and power detection module uses a high-precision power analyzer to measure the voltage, current and power on the low-voltage winding side. The voltage detection range can be selected from 1.5V to 1000V, and the current detection range can be selected from 5mA to 5A. The pulse current and ultrasonic partial discharge detection modules utilize a multi-channel digital partial discharge comprehensive analyzer for detection. A partial discharge measuring instrument input unit is installed at each of the three-phase high-voltage bushings to detect partial discharge on the A / B / C phase high-voltage sides. A current sensor is installed on the grounding wire of both the iron core and the clamping grounding bushing to detect partial discharge at the iron core and clamping points. The pulse current detection frequency band is selected from 40kHz to 300kHz, and the ultrasonic partial discharge detection frequency band is selected from 20kHz to 200kHz. The high-frequency partial discharge detection module uses a high-frequency current sensor combined with a multi-channel digital partial discharge analyzer for detection. One high-frequency current sensor is placed on the grounding wire of both the iron core and the clamping bushing to detect high-frequency partial discharge at the iron core and clamping points. The high-frequency bandwidth is 0.5MHz to 20MHz, the high-frequency dynamic range is 0.001V to 5V, and the measurement time accuracy is 1ns. The oil chromatography online monitoring module has an oil sampling port at the top and bottom of the transformer oil tank, and the oil chromatography analyzer performs online monitoring and analysis of the transformer oil at different locations.

[0028] Step 4: Using the data processing module, the K-Means clustering algorithm is used to classify the multi-parameter feature data corresponding to the burr discharge defects on the transformer core surface under different environmental parameters into four cluster databases according to the temperature and humidity. Combined with the TOPSIS multi-criteria decision algorithm, the correction factor of each feature parameter in each cluster database is calculated based on various detection results, and each feature parameter index weight is assigned. The correction factor and index weight are multiplied and weighted to obtain the fusion weight of each feature parameter. Based on the fusion weight, iterative training is performed to finally obtain the feature database of the burr discharge defects on the transformer core surface. Similarly, following steps one through four, other typical defects such as transformer core loosening, transformer inter-turn short circuit, and transformer clamp bolt loosening and discharge are set up respectively, and finally, feature databases of different defect types are obtained, which together form a transformer defect fault feature database.

[0029] Step 5: Obtain real-time multi-parameter feature data of the transformer to be diagnosed during actual operation. Using the defect diagnosis module, perform similarity analysis between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical defects of transformers under the same environmental parameters in the transformer defect and fault feature database. Diagnose and identify the defect type of the transformer to be diagnosed and output the diagnosis results.

[0030] Example 4 This embodiment provides a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described in Embodiment 2.

[0031] Example 5 This embodiment provides an electronic device, including a memory and a processor: the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, wherein when the computer-executable instructions are executed by the processor, the steps of the method described in embodiment 2 are implemented.

[0032] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0033] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0034] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction set implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0035] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0036] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims of the invention.

[0037] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

Claims

1. A transformer typical defect simulation and diagnosis system based on a fusion algorithm, characterized in that, include: The Euclidean distance calculation module is used to calculate the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault characteristic database. The defect diagnosis module is used to perform similarity analysis using Euclidean distance to determine the defect type of the transformer to be diagnosed.

2. The transformer typical defect simulation and diagnosis system based on fusion algorithm according to claim 1, characterized in that: The transformer defect fault feature database is obtained by iteratively training multi-parameter feature data corresponding to various typical transformer defects under different environmental parameters using a fusion algorithm.

3. The transformer typical defect simulation and diagnosis system based on fusion algorithm according to claim 1, characterized in that: The various typical defects of transformers include transformer core loosening, transformer inter-turn short circuit, transformer core surface burr discharge, and transformer clamp bolt loosening discharge; the multi-parameter characteristic data refers to the set of measured data corresponding to various characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition.

4. The transformer typical defect simulation and diagnosis system based on fusion algorithm according to claim 2, characterized in that: The fusion algorithm includes the K-Means clustering algorithm; The different environmental parameters include humidity and temperature. The multi-parameter feature data corresponding to various typical defects of transformers under different environmental parameters are classified according to the temperature and humidity levels using the K-Means clustering algorithm, resulting in four cluster databases: high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster. In the four cluster databases of high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster, a data point is randomly selected as the initial cluster center for each cluster. The data point includes temperature and humidity. Based on the objective function of the K-Means clustering algorithm, the Euclidean distance from each data point to the initial cluster center of each cluster is calculated, and the data point is assigned to the cluster database corresponding to the initial cluster center with the smallest Euclidean distance. Calculate the average value of the data points for the high temperature and high humidity cluster, the high temperature and low humidity cluster, the low temperature and high humidity cluster, and the low temperature and low humidity cluster respectively, and use them as the new cluster centers; Based on the objective function of the K-Means clustering algorithm, the Euclidean distance from each data point to the new cluster center of each cluster is calculated using the new cluster centers. The data point is then assigned to the cluster database corresponding to the new cluster center with the smallest Euclidean distance. The average value of the data points in the high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster is recalculated and used as the cluster center for each data point in the next iteration. This process continues until the cluster centers no longer change. At this point, each data point is assigned to either the high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, or low temperature and low humidity cluster, thus completing the partitioning of the cluster database.

5. The transformer typical defect simulation and diagnosis system based on fusion algorithm according to claim 4, characterized in that: The fusion algorithm also includes the TOPSIS multi-criteria decision-making algorithm; Each cluster database contains a data sample containing multi-parameter feature data corresponding to a typical defect of a transformer. Combining the TOPSIS multi-criteria decision algorithm, based on the detection results of all feature parameters in each cluster database, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition, correction factors for each feature parameter in the high-temperature high-humidity cluster, high-temperature low-humidity cluster, low-temperature high-humidity cluster, and low-temperature low-humidity cluster are calculated. The mean and standard deviation of vibration signal, port current, voltage and power, pulse current, high-frequency partial discharge and oil chromatographic components in each cluster database are calculated respectively. By calculating the mean and standard deviation of each characteristic parameter in the specific cluster database, the coefficient of variation of each characteristic parameter in the specific cluster database is obtained. Based on the coefficient of variation of each characteristic parameter in the specific cluster database, the correction factor of each characteristic parameter is calculated. Weights are assigned to each characteristic parameter index in the high temperature and high humidity cluster, high temperature and low humidity cluster, low temperature and high humidity cluster, and low temperature and low humidity cluster using the entropy weight method. Organize all data samples from a specific cluster database into a data matrix. X Specific cluster databases contain n There are 10 data samples, each containing all feature parameters; Among them, matrix X This represents the multi-parameter characteristic data corresponding to all typical transformer defects collected under a specific environmental condition; x nm Indicates the first n The first data sample m The numerical values ​​of the characteristic parameters; The values ​​of each feature parameter are normalized and scaled to the [0, 1] interval to calculate the entropy value of each feature parameter, and the difference coefficient of each feature parameter is calculated based on the entropy value of each feature parameter. The difference coefficients of each characteristic parameter are normalized, and the index weights of port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal and oil chromatographic components in each cluster database are calculated respectively. The index weights of port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic components in each cluster database are multiplied by the corresponding correction factors to obtain the fusion weights of port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic components in each cluster database. The feature data in each feature parameter are weighted according to the fusion weight of each feature parameter. The feature data in each feature parameter is multiplied by the corresponding fusion weight of the feature parameter to obtain a weighted dataset. The weighted dataset, environmental parameters, and defect labels are used to train an evaluation model through a regression algorithm to learn the mapping relationship between the feature data in each feature parameter and the defect type. The evaluation model is used to predict the predicted value for each data sample. The error between the predicted value and the actual value of the defect label is calculated. If the error is greater than A, the data sample is marked as erroneous data and removed. After removing erroneous data, a new dataset is obtained. The fusion weight of each feature parameter is recalculated, and the above process is repeated. The training is iterated until the maximum number of iterations is reached, and finally, a transformer defect fault feature database is established.

6. The transformer typical defect simulation and diagnosis system based on fusion algorithm according to claim 5, characterized in that, By calculating the Euclidean distance between the real-time multi-parameter feature data of the transformer under diagnosis and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database, similarity analysis is performed using the Euclidean distance to obtain the defect type of the transformer under diagnosis. Specifically, this is used for: Calculate the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect and fault characteristic database. d(x,y) The calculation formula is: in, x j This represents the first of the real-time multi-parameter characteristic data of the transformer to be diagnosed. j Feature data in the feature parameters; y j This represents the first of the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database. j Feature data in the feature parameters; m It represents seven characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition. European distance d(x,y) The smaller the value, the higher the similarity. The Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database is compared, and the Euclidean distance is selected. d(x,y) The smallest typical transformer defect is used as the defect type of the transformer to be diagnosed.

7. A method for simulating and diagnosing typical transformer defects based on a fusion algorithm, characterized in that, Includes the following steps: The Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical defects of transformers under the same environmental parameters in the transformer defect and fault characteristic database is calculated. The type of defect in the transformer to be diagnosed is obtained by similarity analysis using Euclidean distance.

8. The method for simulating and diagnosing typical transformer defects based on a fusion algorithm according to claim 7, characterized in that: The transformer defect fault feature database is obtained by iteratively training multi-parameter feature data corresponding to various typical transformer defects under different environmental parameters using a fusion algorithm.

9. The method for simulating and diagnosing typical transformer defects based on a fusion algorithm according to claim 7, characterized in that, By calculating the Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database, similarity analysis is performed using the Euclidean distance to obtain the defect type of the transformer to be diagnosed, including: Calculate the Euclidean distance between the real-time multi-parameter characteristic data of the transformer to be diagnosed and the multi-parameter characteristic data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect and fault characteristic database. d(x,y) The calculation formula is: in, x j This represents the first of the real-time multi-parameter characteristic data of the transformer to be diagnosed. j Feature data in the feature parameters; y j This represents the first of the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database. j Feature data in the feature parameters; m It represents seven characteristic parameters, including port voltage, port current, reactive power loss, pulse current, high-frequency partial discharge, vibration signal, and oil chromatographic composition. European distance d(x,y) The smaller the value, the higher the similarity. The Euclidean distance between the real-time multi-parameter feature data of the transformer to be diagnosed and the multi-parameter feature data corresponding to various typical transformer defects under the same environmental parameters in the transformer defect fault feature database is compared, and the Euclidean distance is selected. d(x,y) The smallest typical transformer defect is used as the defect type of the transformer to be diagnosed.

10. A computer storage medium, wherein the computer-readable storage medium stores a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the transformer typical defect simulation and diagnosis method based on the fusion algorithm as described in any one of claims 7-9.

Images