Current transformer defect detection method and system

Abstract

The present invention relates to the field of transformer detection, and more specifically, to a method and system for detecting defects in a current transformer. The method comprises: obtaining corresponding feature vectors based on different detection methods, training a preset network by collecting multi-dimensional detection data of the current transformer in history; calculating the inspection consistency of any two dimensions; for any detection environment, calculating the comprehensive detection result of the current transformer based on the inspection consistency, traversing to obtain the comprehensive detection results of all detection environments, weighting and adding the comprehensive detection results as the final detection result of the current transformer, and completing the detection. Through the technical solution of the present invention, the accuracy of the current transformer detection result can be improved, ensuring the normal operation of the current transformer.

Description

Technical Field

[0001] The present invention relates to the field of transformer detection, and more specifically, to a current transformer defect detection method and system. Background Art

[0002] Current transformers are key devices for measurement and protection in power systems. They convert high-voltage currents into low-voltage currents, allowing measuring instruments, protection devices, and control equipment to operate at safe voltage levels. The performance of current transformers directly affects the reliability and safety of power systems. Current transformers will have different operating environments during long-term operation, and may have defects for different reasons, such as insulation aging, winding short circuits, core overheating, etc. These defects may lead to measurement errors, equipment damage, and even power accidents, posing a serious threat to the stable operation of power systems.

[0003] The existing Chinese patent application document with publication number CN112017173A discloses a method for detecting defects in power equipment based on a target detection network and structured positioning. The method includes: marking the equipment category and component structure category of each power equipment infrared image, generating a data set and preprocessing it; inputting the preprocessed data set into a target detection model to detect the target power equipment type; using a structured positioning algorithm to locate the components of the power equipment; and detecting defects in the power equipment based on the temperature data of the infrared images of each power equipment and the infrared diagnostic rules of the energized equipment.

[0004] However, the application scenarios of current transformers are usually flexible and changeable. The above scheme only uses a single thermal imaging detection method to detect defects in power equipment, which may result in different detection results in different application scenarios, thereby leading to inaccurate defect detection results. Summary of the invention

[0005] In order to solve the problem of inaccurate defect detection results, the present invention proposes a current transformer defect detection method and system.

[0006] In a first aspect, the present invention discloses a method for detecting defects in a current transformer, comprising: obtaining corresponding feature vectors based on different detection methods, training a preset network by collecting multi-dimensional detection data of the current transformer in history, wherein one dimension corresponds to a preset network, and the detection data includes a feature vector and a detection result; calculating the inspection consistency of any two dimensions; for any detection environment, calculating the comprehensive detection result of the current transformer based on the inspection consistency, traversing to obtain the comprehensive detection results of all detection environments, weighting and adding the comprehensive detection results as the final detection result of the current transformer, and completing the detection.

[0007] On the premise of ensuring the consistency of the detection environment, the preset network model can be used to obtain the judgment results of current transformer defects from various dimensions, revealing the behavioral characteristics of the current transformer in different detection dimensions. By calculating the inspection consistency of different dimensions, the correlation between different detection methods in different dimensions can be obtained. Then, by integrating the results of each detection environment and adding them with weights, a final detection result that comprehensively reflects the status of the current transformer is obtained. This not only improves the accuracy of detection, but also helps to effectively prevent potential power system failures, thereby ensuring the stability and safety of power supply.

[0008] Preferably, the preset network is a BP network, which includes an input layer, a hidden layer and an output layer. The input layer is used to receive a feature vector of any dimension. After the hidden layer extracts features from the input information, the extracted features are input into the output layer to output a predicted value of the detection result of any dimension.

[0009] Preferably, the training process of the preset network includes: taking the feature vector of any dimension as input information, and taking the true value of the detection result of any dimension as a label to obtain a set of training data; inputting the input information in the training data into the preset network to obtain an output result; calculating the loss value of the preset network through the output result and the label, back-propagating the error signal according to the loss value, updating the network parameters of the prediction model to reduce the loss value; iteratively updating the network parameters of the prediction model, and stopping the update when the prediction model reaches the set maximum number of training times or the network loss value is less than the set loss value to obtain a trained preset network.

[0010] Preferably, the consistency check satisfies the relationship:

[0011] , Representation Dimension and dimensions The consistency of the test, and , and Respectively represent dimensions The test results are qualified and dimension The test result is qualified. and Respectively represent dimensions The test results are unqualified and dimension The test result is unqualified. and They respectively indicate that the true value of the final test result of the current transformer is qualified and the true value of the final test result is unqualified. Indicates quantity.

[0012] Based on the quantitative relationship between the test results of the two dimensions and the true value of the final test result of the current transformer, that is, combining the mutual relationship between different dimensions and the comparison of the true value, it is possible to more accurately evaluate and compare the consistency of the test results of different dimensions, effectively detect and eliminate possible systematic errors or deviations, and improve the reliability and accuracy of the overall test results.

[0013] Preferably, the verification consistency also includes: taking any two dimensions as the first dimension and the second dimension, respectively calculating the similarity between the first dimension and the second dimension, the similarity between the first dimension and the true value of the final detection result, and the similarity between the second dimension and the true value of the final detection result, and taking the mean of the similarities as the verification consistency of the first dimension and the second dimension.

[0014] By comparing any two dimensions with the final result, calculating their similarity, and measuring the coordination between dimensions by the mean, it is ensured that each dimension can collaboratively reflect the actual detection situation, which also helps to discover potential sources of errors.

[0015] Preferably, the comprehensive detection result satisfies the relationship:

[0016] , Indicates the comprehensive test results. Represents the total number of dimensions, Representation Dimension and dimensions The consistency of the test, Representation Dimension The output of the preset network.

[0017] Not only does it take into account the consistency between dimensions, but it also combines the relative importance of each dimension in the overall detection, thereby avoiding the deviation or instability factors that may be caused by a certain dimension. This can more accurately reflect the overall performance of the system, ensure that more consistent and reliable results can be obtained in multiple dimensions, and improve the credibility and efficiency of the detection system.

[0018] Preferably, the comprehensive detection result also satisfies the relationship:

[0019] , Indicates the comprehensive test results. Represents the total number of dimensions, Representation Dimension and dimensions The consistency of the test, Representation Dimension The output of the preset network, Representation Dimension The accuracy of the preset network.

[0020] In multi-dimensional detection, some dimensions may have lower accuracy for some reasons. By introducing the accuracy of the preset network as the weight coefficient, the impact of these dimensions can be reduced during comprehensive detection, so that the overall detection results can more truly reflect the actual contribution and performance of each dimension, and a more reasonable balance can be achieved between multiple dimensions.

[0021] In a second aspect, the present invention discloses a current transformer defect detection system, comprising: a processor; and a memory, wherein the memory stores computer instructions, and when the computer instructions are executed by the processor, the system executes the above-mentioned current transformer defect detection method.

[0022] Beneficial effects of the present invention:

[0023] The present invention can more accurately identify potential defects of current transformers through comprehensive analysis of multi-dimensional feature vectors combined with consistency checks between different detection dimensions. Specifically, based on the training of historical detection data, BP neural network is used for feature extraction and prediction. By calculating the consistency and similarity of the inspections between different dimensions, the results of multiple detection dimensions can be effectively integrated, thereby improving the accuracy and robustness of the detection. By weighted fusion of the detection results of each dimension and considering the accuracy of each dimension, the overall health status of the current transformer can be comprehensively evaluated, avoiding possible misjudgments or omissions in a single dimension, and ultimately achieving more comprehensive and accurate detection of current transformer faults. This method not only improves the detection accuracy, but also improves the reliability of current transformer defect diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0025] Figure 1 It is a flow chart of a current transformer defect detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0027] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0028] The present invention provides a method for detecting defects in a current transformer. Figure 1 As shown, the current transformer defect detection method includes steps S1 to S3, which are described in detail below.

[0029] S1, obtain corresponding feature vectors based on different detection methods, and train the preset network by collecting multi-dimensional detection data of the current transformer in history.

[0030] It should be noted that when testing the current transformer, special test instruments and equipment are required, such as current transformer test equipment, voltmeter, ammeter, load resistor, etc. At the same time, attention should be paid to the stability and accuracy of the test environment to avoid the influence of external factors on the test results. The test methods of current transformer mainly include rated parameter test, no-load test, ratio test, load characteristic test, temperature rise test, insulation test and impedance test. Through these tests, the performance and accuracy of the current transformer can be evaluated to ensure its reliability and stability in practical applications.

[0031] Exemplary defect detection methods include the direct current method: by applying direct current, observing the ratio of the output of the current transformer to the input current, judging whether the ratio of the current transformer meets the standard. AC test method: using AC signals for testing, detecting the phase difference and amplitude of the current transformer, and evaluating its accuracy and linearity. Frequency response analysis: by changing the input frequency, analyzing the frequency response of the current transformer, identifying possible resonance or other changes in frequency characteristics. Thermal imaging detection: using infrared thermal imaging technology to detect the temperature distribution of the current transformer, identifying possible overheating or poor contact problems. Ultrasonic detection: using ultrasonic equipment to check for structural defects inside the current transformer, such as aging or cracks in the insulating material. Insulation resistance test: detecting the insulation resistance of the current transformer to evaluate the performance of the insulating material and potential leakage risks.

[0032] In one embodiment, characteristic vectors are obtained according to different detection methods. For the same current transformer, one method corresponds to one characteristic vector.

[0033] Collect multi-dimensional detection data of multiple current transformers in history, the detection data includes feature vectors and detection results, the detection result is qualified or unqualified, and one dimension corresponds to one detection method.

[0034] The preset network is trained through multi-dimensional detection data, and one dimension corresponds to one preset network. The preset network is a BP network, which includes an input layer, a hidden layer and an output layer. The input layer is used to receive the feature vector of any dimension. After the hidden layer extracts the features of the input information, the extracted features are input to the output layer to output the predicted value of the detection result of any dimension.

[0035] The training process of the preset network includes: taking the feature vector of any dimension as input information, and taking the true value of the detection result of any dimension as a label, to obtain a set of training data; inputting the input information in the training data into the preset network to obtain the output result; calculating the loss value of the preset network through the output result and the label, back-propagating the error signal according to the loss value, updating the network parameters of the prediction model, and making the loss value smaller; iteratively updating the network parameters of the prediction model, and when the prediction model reaches the set maximum number of training times or the network loss value is less than the set loss value, stop updating to obtain a trained preset network. Exemplarily, when the number of training times of the preset network reaches 200 times or the network loss value is less than 0.0001, stop updating.

[0036] At this point, the trained preset network corresponding to each dimension can be obtained.

[0037] S2, calculate the test consistency of any two dimensions.

[0038] It should be noted that for the same current transformer, the test results obtained by different test methods under the same test environment may be different, because each method has its own unique measurement principle and scope of application. The DC current method focuses on detecting the proportional characteristics of the transformer by applying DC current. The AC test method uses AC signals to evaluate the phase difference and amplitude of the transformer to determine its accuracy and linearity. Frequency response analysis identifies changes in the frequency characteristics of the transformer by changing the input frequency. Thermal imaging detection uses infrared technology to detect the temperature distribution of the transformer. Ultrasonic testing uses ultrasonic equipment to check for structural defects inside the transformer. Insulation resistance testing evaluates the performance of the insulating material and potential leakage risks by measuring the insulation resistance of the transformer.

[0039] Since these methods focus on different parameters and detection emphases, they may give different results in practical applications. Therefore, it is necessary to calculate the consistency of the detection results of two different detection methods.

[0040] In one embodiment, the consistency check satisfies the relationship:

[0041] , Representation Dimension and dimensions The consistency of the test, and , and Respectively represent dimensions The test results are qualified and dimension The test result is qualified. and Respectively represent dimensions The test results are unqualified and dimension The test result is unqualified. and They respectively indicate that the true value of the final test result of the current transformer is qualified and the true value of the final test result is unqualified. Indicates quantity.

[0042] Representation Dimension Test results, dimensions The sum of the number of cases where the test result of and the true value of the final test result of the current transformer are consistent; Represents the number of all cases, i.e., the dimension Test results, dimensions There is a situation where any one of the detection results and the true value of the final detection result of the current transformer is different from the other two detection results and there is a situation where the three detection results are the same.

[0043] The greater the inspection consistency, the higher the consistency of the inspection results obtained by the inspection methods in the two dimensions. When inspecting a new current transformer, the inspection results obtained by the inspection method with high inspection consistency have high credibility.

[0044] In one embodiment, verifying consistency also includes: taking any two dimensions as the first dimension and the second dimension, respectively calculating the similarity between the first dimension and the second dimension, the similarity between the first dimension and the true value of the final detection result, and the similarity between the second dimension and the true value of the final detection result, and taking the average of the similarities as the verification consistency of the first dimension and the second dimension.

[0045] It should be noted that when calculating the consistency of the test, multiple samples are required. For example, there are five current transformers, that is, there are five feature vectors in the first dimension to construct the first dimension set, and there are five feature vectors in the second dimension to construct the second dimension set, which will also correspond to five true value sequences of the final test results. The true value sequence is a 01 sequence of qualified or unqualified, 0 means unqualified, and 1 means qualified.

[0046] S3, for any detection environment, the comprehensive detection result of the current transformer is calculated based on the verification consistency, the comprehensive detection results of all detection environments are traversed to obtain, the comprehensive detection results are weighted and added as the final detection result of the current transformer, and the detection is completed.

[0047] It should be noted that the detection environment will affect the defect detection results of the current transformer, because the test of the current transformer needs to be carried out within a specific temperature range to ensure the accuracy of the test results. For example, the test environment temperature is usually required to be between 5°C and 40°C. Beyond this range, the effect of temperature on the test results may become significant, resulting in increased errors. There should be no obvious AC or DC external electromagnetic field influence in the test site, as this may interfere with the test results and affect the accurate evaluation of the current transformer performance. Humidity and harsh environmental conditions may have a negative impact on the performance of the current transformer. When the current transformer is error-detected under low voltage, due to the presence of capacitance between the primary and secondary windings, there will be leakage current flowing from the high-voltage primary winding to the low-voltage secondary winding. Therefore, studying the performance of the current transformer under different detection environments is crucial to ensure its accuracy and reliability in practical applications.

[0048] In one embodiment, for any detection environment, the comprehensive detection result of the current transformer satisfies the relationship:

[0049] , Indicates the comprehensive test results. Represents the total number of dimensions, Representation Dimension and dimensions The consistency of the test, Representation Dimension The output of the preset network.

[0050] Representation Dimension Divide the dimension The cumulative value of the consistency of all dimensions except Represents the cumulative value of all dimensions.

[0051] It should be noted that in the relationship of the comprehensive test results, the dimension Although the upper limit of , but there are discontinuities, which are not reflected in the formula here. For example, there are 5 dimensions in total. Take 1, Take 2, 3, 4 and 5 respectively and add them up. However, when Take 2, Take 1, 3, 4 and 5 respectively.

[0052] The comprehensive detection results of all detection environments are traversed and the results of weighting the comprehensive detection results are added as the final detection result of the current transformer. The weighting is a well-known technology for those skilled in the art. Since most of the current transformers actually work in the same detection environment, the weights in the commonly used detection environment can be set by those skilled in the art to be larger than those in other detection environments to reflect the actual working conditions of the current transformer.

[0053] In one embodiment, the comprehensive detection result also satisfies the relationship:

[0054] , Indicates the comprehensive test results. Represents the total number of dimensions, Representation Dimension and dimensions The consistency of the test, Representation Dimension The output of the preset network, Representation Dimension The accuracy of the preset network.

[0055] There will be an accuracy rate in the preset network corresponding to each dimension, and the accuracy rate is a technology well known to those skilled in the art.

[0056] In one embodiment, a Gaussian kernel function is set according to the detection environment of the current transformer to determine the importance of the detection results of different detection environments, and the importance satisfies the relationship:

[0057] , Indicates the detection environment The importance of the test results, The most suitable detection environment for current transformers is the application environment of current transformers. represents the variance, Represents an exponential function.

[0058] Exemplarily, the variance is set to 1, that is, a Gaussian distribution with a variance of 1 is selected to calculate the importance.

[0059] The closer the test environment is to the application environment, the more important the test results in the test environment are.

[0060] Importance is used as weight to assign weights to the comprehensive test results to complete the test.

[0061] An embodiment of the present invention further discloses a current transformer defect detection system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a current transformer defect detection method according to the present invention is implemented.

[0062] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0063] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module or both. Any such computer storage medium may be part of a device or accessible or connectable to a device.

[0064] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0065] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A current transformer defect detection method, characterized in that: include: Based on different detection methods, corresponding feature vectors are obtained, and a preset network is trained by collecting multi-dimensional detection data of the current transformer in history, wherein one dimension corresponds to one preset network, and the detection data includes feature vectors and detection results; Calculate the test consistency of any two dimensions; For any detection environment, the comprehensive detection result of the current transformer is calculated based on the verification consistency, the comprehensive detection results of all detection environments are traversed to obtain, the comprehensive detection results are weighted and added as the final detection result of the current transformer, and the detection is completed; The consistency test satisfies the relationship: , Representation Dimension and dimensions The consistency of the test, and , and Respectively represent dimensions The test results are qualified and dimension The test result is qualified. and Respectively represent dimensions The test results are unqualified and dimension The test result is unqualified. and They respectively indicate that the true value of the final test result of the current transformer is qualified and the true value of the final test result is unqualified. Indicates quantity; Representation Dimension Test results, dimensions The sum of the number of cases where the test result of and the true value of the final test result of the current transformer are consistent; Represents the number of all cases, i.e., the dimension Test results, dimensions There is a situation where any one of the detection results and the true value of the final detection result of the current transformer is different from the other two detection results and there is a situation where the three detection results are the same.

2. The current transformer defect detection method according to claim 1, characterized in that: The preset network is a BP network, which includes an input layer, a hidden layer and an output layer. The input layer is used to receive a feature vector of any dimension. After the hidden layer extracts features from the input information, the extracted features are input to the output layer to output a predicted value of the detection result of any dimension.

3. The current transformer defect detection method according to claim 1, characterized in that: The training process of the preset network includes: Take the feature vector of any dimension as input information, and take the true value of the detection result of any dimension as the label to obtain a set of training data; Input the input information in the training data into the preset network to obtain the output result; The loss value of the preset network is calculated by outputting the results and labels, and the error signal is back-propagated according to the loss value to update the network parameters of the prediction model to reduce the loss value. Iteratively update the network parameters of the prediction model. When the prediction model reaches the set maximum number of training times or the network loss value is less than the set loss value, stop updating and obtain a trained preset network.

4. The current transformer defect detection method according to claim 1, characterized in that: The comprehensive test results satisfy the relationship: , Indicates the comprehensive test results. Represents the total number of dimensions, Representation Dimension and dimensions The consistency of the test, Representation Dimension The output of the preset network.

5. The current transformer defect detection method according to claim 1, characterized in that: The comprehensive test results also satisfy the relationship: , Indicates the comprehensive test results. Represents the total number of dimensions, Representation Dimension and dimensions The consistency of the test, Representation Dimension The output of the preset network, Representation Dimension The accuracy of the preset network.

6. Current transformer defect detection system, characterized in that: include: Processor; and A memory storing computer instructions, wherein when the computer instructions are executed by a processor, the system executes the current transformer defect detection method according to any one of claims 1 to 5.

Images