State judgment method for zero sequence mutual inductor

Abstract

The invention discloses a state judgment method for a zero-sequence mutual inductor, and the method comprises the following steps: obtaining real-time operation parameters of the zero-sequence mutual inductor according to a preset period or an external instruction, the operation parameters at least comprising a zero-sequence current, a zero-sequence voltage, an environment temperature and an impedance parameter of a mutual inductor winding; preprocessing the operation parameters, and comparing a preprocessing result with pre-stored reference parameters to generate a deviation feature vector; inputting the deviation feature vector into a pre-trained state judgment model, and outputting a state judgment result of the zero sequence transformer, the state judgment result including a normal state, an early warning state and a fault state; and executing corresponding control operation according to the state judgment result.

Description

Technical Field

[0001] The present invention relates to the technical field of mutual inductor judgment, and particularly to a method for judging the state of a zero-sequence mutual inductor. Background Art

[0002] With the continuous development of the power system and the improvement of the intelligent level, the zero-sequence mutual inductor, as a key protection device in the power system, the accurate judgment of its operating state is of great significance for ensuring the safe and stable operation of the power system. The zero-sequence mutual inductor is mainly used to detect the zero-sequence current and zero-sequence voltage in the power system, so as to achieve a rapid response and protection for system grounding faults.

[0003] At present, the current detection method of the zero-sequence mutual inductor is to press the test button on the product monthly to simulate a leakage signal, and take whether the protection circuit breaker trips as the judgment basis to judge whether there is a failure phenomenon in the product. These methods have the following problems and deficiencies:

[0004] Low efficiency of manual detection: The traditional manual detection method requires professional personnel to conduct on-site detection regularly, which is not only time-consuming and laborious, but also has a limited detection frequency and is difficult to achieve real-time monitoring.

[0005] Single detection means: The existing detection methods usually use a single parameter threshold judgment, lacking comprehensive analysis of multi-dimensional parameters, which is prone to misjudgment or missed judgment.

[0006] Lack of intelligence: The traditional method lacks intelligent analysis means and cannot perform automatic state judgment based on historical data and real-time data, making it difficult to adapt to the complex and changeable operating environment of the power system.

[0007] In addition, when dealing with the aging warning and fault diagnosis of the zero-sequence mutual inductor, the existing state judgment methods often rely on expert experience and lack scientific quantitative analysis methods. This not only increases the maintenance cost, but also may lead to delayed discovery of equipment failures, thus affecting the safe operation of the power system. Summary of the Invention

[0008] In view of the above-mentioned prior art, the present invention aims to provide a method for judging the state of a zero-sequence mutual inductor, mainly solving the technical problems existing in the above background art.

[0009] To achieve the above object, the technical solution of the embodiment of the present invention is realized as follows:

[0010] A method for judging the state of a zero-sequence mutual inductor, the method comprising the following steps:

[0011] According to a preset period or an external instruction, obtain the real-time operating parameters of the zero-sequence mutual inductor, where the operating parameters at least include zero-sequence current, zero-sequence voltage, ambient temperature, and impedance parameters of the mutual inductor winding;

[0012] Preprocess the operating parameters, compare the preprocessing result with the pre-stored reference parameters, and generate a deviation feature vector;

[0013] Input the deviation feature vector into a pre-trained state judgment model to output the state judgment result of the zero-sequence current transformer, where the state judgment result includes a normal state, a warning state, and a fault state;

[0014] Execute corresponding control operations according to the state judgment result.

[0015] Optionally, obtain the real-time operating parameters of the zero-sequence current transformer according to a preset period or an external instruction, specifically including: generating a simulated leakage signal of a fixed magnitude for the zero-sequence coil according to a preset period or an external instruction, and obtaining the real-time operating parameters of the zero-sequence current transformer based on the simulated leakage signal.

[0016] Optionally, preprocess the operating parameters, compare the preprocessing result with the pre-stored reference parameters, and generate a deviation feature vector, specifically including:

[0017] Perform data cleaning on the operating parameters to remove outliers and noise data;

[0018] Perform normalization processing on the cleaned data to obtain real-time operating parameters;

[0019] Compare the real-time operating parameters with the pre-stored reference parameters to generate a deviation feature vector.

[0020] Optionally, compare the real-time operating parameters with the pre-stored reference parameters to generate a deviation feature vector, specifically including:

[0021] During the factory calibration stage of the zero-sequence current transformer, collect its zero-sequence current I ref , zero-sequence voltage U ref , and zero-sequence impedance Z ref ;

[0022] Calculate the relative current deviation ΔI, relative voltage deviation ΔU, and relative impedance deviation ΔZ respectively through the following formula:

[0023]

[0024]

[0025] where I A is the real-time leakage current, U A is the real-time leakage voltage, and Z A is the real-time impedance;

[0026] Combine the relative current deviation ΔI, the relative voltage deviation ΔU, the relative impedance deviation ΔZ, and the ambient temperature to form a deviation feature vector D = {ΔI i , ΔU i , ΔZ i , T}.

[0027] Optionally, train the state judgment model, specifically including:

[0028] Collect a historical deviation data set V = {(d 1 , y 1 ), (d 2 , y 2 ),...,(d n , y n )}, where d n is the nth historical deviation feature vector, and y n ∈ {normal, aging warning, fault} is the state label;

[0029] Build a normal state SVM classifier and an aging warning SVM classifier. With normal state samples as the positive class and aging warning and fault state samples as the negative class, train the normal state SVM classifier through the optimization objective of maximizing the classification margin to obtain the hyperplane H 1 ; With aging warning samples as the positive class and fault state samples as the negative class, train the aging warning SVM classifier through the optimization objective of maximizing the classification margin to obtain the hyperplane H 2 .

[0030] Optionally, input the deviation feature vector into the pre-trained state judgment model to output the state judgment result of the zero-sequence current transformer, specifically including: input the deviation feature vector D into the normal state SVM classifier to obtain a first result value. If the first result value is less than the first threshold, it indicates that the zero-sequence current transformer is in the normal state; if the first result value is greater than the first threshold, input the deviation feature vector D into the aging warning SVM classifier to obtain a second result value. If the second result value is less than the second threshold, it indicates that the zero-sequence current transformer is in the aging state; if the second result value is greater than the second threshold, it indicates that the zero-sequence current transformer is in the fault state.

[0031] Optionally, set the first threshold and the second threshold, specifically including: substitute the historical deviation feature vector d n into the decision function f 1 (d 1 ) of the hyperplane H n and the decision function f 2 of the hyperplane H 2 (d n) First, obtain the first decision value and the second decision value respectively. Aggregate multiple first decision values into a first array, and aggregate multiple second decision values into a second array. Calculate the 99% quantile of the first array as the first threshold, and calculate the 50% quantile of the second array as the second threshold.

[0032] Optionally, when the zero-sequence current transformer is in a normal state, generate a self-inspection qualified report;

[0033] When the zero-sequence current transformer is in an aging state, generate a marked transformer identification;

[0034] When the zero-sequence current transformer is in a fault state, trigger a blocking signal to disconnect the output circuit of the zero-sequence current transformer.

[0035] The beneficial effects of the present invention are as follows: The state judgment method for the zero-sequence current transformer provided by this application significantly improves the detection efficiency, avoids the cumbersome operations and potential risks of traditional manual detection, and enables the state of the zero-sequence current transformer to be quickly evaluated in a short time. By using a state judgment model, it greatly enhances the accuracy of detection, reduces the possibility of misjudgment and missed judgment, and ensures the reliability of the state judgment of the zero-sequence current transformer. In addition, the present invention realizes real-time monitoring of the zero-sequence current transformer, can timely detect the abnormal state of the equipment, give early warnings and take corresponding measures, effectively avoid power outages and maintenance costs caused by equipment failures, and reduce the overall maintenance cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic flow chart of the state judgment method for the zero-sequence current transformer in the embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] The technical solutions of the present invention will be further described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. In the following description, the expression "some embodiments" is used, which describes a subset of all possible embodiments. However, it should be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0038] In the following description, a large number of specific details are given to provide a more thorough understanding of the present invention. However, it is obvious to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, some well-known technical features are not described to avoid confusion with the present invention.

[0039] It should be understood that the present invention can be implemented in different forms and should not be construed as limited to the embodiments presented herein. On the contrary, providing these embodiments will make the disclosure thorough and complete, and will fully convey the scope of the present invention to those skilled in the art. And the purpose of the terms used herein is only to describe specific embodiments and is not a limitation of the present invention. As used herein, the singular forms "a", "an" and "the" are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the terms "comprising" and / or "including", when used in this specification, determine the presence of the stated features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups. As used herein, the term "and / or" includes any and all combinations of the related listed items.

[0040] It should be further noted that when an element is referred to as "fixed to" another element, it can be directly on the other element or there can also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "inner", "outer", "left", "right" and similar expressions used herein are for illustrative purposes only and do not represent the only embodiments.

[0041] To thoroughly understand the present invention, detailed structures will be presented in the following description to illustrate the technical solutions proposed by the present invention. The optional embodiments of the present invention are described in detail below. However, in addition to these detailed descriptions, the present invention can also have other embodiments.

[0042] Please refer to the attached Figure 1 , this application provides a method for judging the state of a zero-sequence current transformer, and the method includes the following steps:

[0043] S1. Obtain the real-time operating parameters of the zero-sequence current transformer according to a preset period or an external instruction, and the operating parameters at least include zero-sequence current, zero-sequence voltage, ambient temperature and impedance parameters of the transformer winding;

[0044] S2. Preprocess the operating parameters, and compare the preprocessing result with the pre-stored reference parameters to generate a deviation feature vector;

[0045] S3. Input the deviation feature vector into a pre-trained state judgment model, and output the state judgment result of the zero-sequence current transformer, and the state judgment result includes a normal state, a warning state and a fault state;

[0046] S4. Execute corresponding control operations according to the state judgment result.

[0047] In an alternative embodiment, in step S1, the preset period or external instruction includes the following operations. The first is to issue a gateway instruction, collect the gateway instruction through the 485 module for external communication, and drive the zero-sequence current transformer to perform a self-check operation through the gateway instruction. The second is to output a self-check instruction through the button signal on the body, and the zero-sequence current transformer performs a self-check operation based on this self-check instruction.

[0048] When performing the self-check operation, the external MCU will generate a fixed-size simulated leakage signal for the zero-sequence coil. Based on the simulated leakage signal, the real-time operating parameters of the zero-sequence current transformer at this time are recorded, such as zero-sequence current, zero-sequence voltage, ambient temperature, and impedance parameters of the transformer winding.

[0049] In an alternative embodiment, in step S2, the operating parameters are preprocessed, and the preprocessing result is compared with the pre-stored reference parameters to generate a deviation feature vector. Specifically, it includes:

[0050] Perform data cleaning on the operating parameters to remove outliers and noise data;

[0051] Perform normalization processing on the cleaned data to obtain real-time operating parameters;

[0052] Compare the real-time operating parameters with the pre-stored reference parameters to generate a deviation feature vector.

[0053] In the process of preprocessing the operating parameters of the zero-sequence current transformer, it is first necessary to perform data cleaning on the collected operating parameters. The main purpose of this process is to remove outliers and noise data in them, so as to ensure the accuracy and reliability of subsequent analysis. Outliers may be caused by instantaneous errors of sensors, external interference, or other unforeseen factors. If these outliers are not removed, they may have a greater impact on the subsequent generation of deviation feature vectors, thereby affecting the accuracy of state judgment.

[0054] Noise data is usually caused by the accuracy limitation of measurement equipment or environmental factors. These noise data will mask the true operating state information. Through data cleaning, the interference of noise can be effectively reduced, making the subsequent analysis more accurate. After data cleaning is completed, the next step is to perform normalization processing on the cleaned data, that is, to scale the data according to a certain ratio so that it falls within a specific range, such as [0, 1] or [-1, 1]. The main purpose of normalization processing is to eliminate the dimensional difference between different parameters, so that each parameter is comparable in subsequent analysis.

[0055] Since the zero-sequence current, zero-sequence voltage, ambient temperature, and impedance parameters of the transformer winding have different dimensions and numerical ranges, direct comparison and analysis may lead to deviations in the generation of deviation characteristic vectors. Through normalization, these parameters can be unified to the same scale, thus more accurately reflecting the operating state of the zero-sequence current transformer.

[0056] The real-time operating parameters obtained after normalization can more accurately reflect the actual operating conditions of the zero-sequence current transformer, providing a reliable data basis for generating deviation characteristic vectors by comparing with pre-stored reference parameters. Through the above preprocessing process, the accuracy and reliability of the zero-sequence current transformer state judgment can be effectively improved, providing strong support for subsequent state classification and determination.

[0057] Furthermore, comparing the real-time operating parameters with the pre-stored reference parameters to generate deviation characteristic vectors, the specific steps for generating deviation characteristic vectors are as follows:

[0058] During the factory calibration stage of the zero-sequence current transformer, collect its zero-sequence current I ref , zero-sequence voltage U ref , and zero-sequence impedance Z ref ;

[0059] Calculate the current relative deviation ΔI, voltage relative deviation ΔU, and impedance relative deviation ΔZ respectively through the following formula:

[0060]

[0061] where I A is the real-time leakage current, U A is the real-time leakage voltage, and Z A is the real-time impedance;

[0062] Combine the current relative deviation ΔI, voltage relative deviation ΔU, impedance relative deviation ΔZ, and ambient temperature to form a deviation characteristic vector D = {ΔI i , ΔU i , ΔZ i , T}.

[0063] By comparing the real-time operating parameters with the reference parameters, the difference between the current operating state of the zero-sequence current transformer and the standard state can be quantified, thus providing a strong basis for subsequent state judgment. When the zero-sequence current transformer is in normal operation, warning state, and fault state, its zero-sequence current, zero-sequence voltage, ambient temperature, and impedance parameters of the transformer winding will show different characteristics. By comparing the real-time collected parameters with the pre-stored reference parameters, the deviation degree of each parameter can be calculated, and then a deviation characteristic vector that can reflect the state of the zero-sequence current transformer can be generated.

[0064] In terms of the effect, this comparison process can effectively capture the subtle changes in the operating state of the zero-sequence current transformer. For example, when abnormal fluctuations occur in the zero-sequence current or zero-sequence voltage, these abnormalities can be detected in a timely manner and corresponding deviation feature vectors can be generated through comparison with the reference parameters. These deviation feature vectors not only contain the deviation information of each parameter, but also can reflect the overall operating condition of the zero-sequence current transformer through a combination method. In subsequent state judgment, the deviation feature vectors will serve as important input data to help the state judgment model accurately distinguish between normal state, early warning state, and fault state. Therefore, comparing the real-time operating parameters with the pre-stored reference parameters and generating deviation feature vectors is a key link in realizing the accurate judgment of the state of the zero-sequence current transformer, which plays an important role in improving the reliability and safety of equipment operation.

[0065] In step S3, inputting the deviation feature vector into the pre-trained state judgment model is one of the core steps of the present invention to realize the accurate judgment of the state of the zero-sequence current transformer. The state judgment model is pre-trained based on a large amount of historical data and advanced machine learning algorithms, and it can effectively identify the deviation feature vector patterns in different states. By inputting the deviation feature vector obtained in real time into the model, the model will automatically perform complex calculations and analyses, and then output the state judgment result of the zero-sequence current transformer.

[0066] This result includes normal state, early warning state, and fault state, comprehensively covering various working conditions that the zero-sequence current transformer may be in, providing a direct basis for subsequent control operations. The model comprehensively considers information such as current relative deviation, voltage relative deviation, impedance relative deviation, and ambient temperature contained in the deviation feature vector, and matches and compares them with various state features learned during the pre-training stage. When the similarity between the deviation feature vector and the feature pattern of the normal state is relatively high, the model will judge that the zero-sequence current transformer is in the normal state; if the deviation feature vector shows a certain degree of abnormality but has not reached the fault level, the model will judge it as the early warning state, prompting relevant personnel to pay attention to the equipment condition; and when the deviation feature vector highly coincides with the feature of the fault state, the model will accurately judge the fault state so as to take timely measures to avoid potential safety risks and equipment damage. This model-based state judgment method not only improves the accuracy and reliability of judgment, but also realizes the real-time monitoring and dynamic evaluation of the state of the zero-sequence current transformer. Compared with traditional manual detection methods, it greatly improves the detection efficiency and intelligent level, effectively solves the problems of cumbersome operation, low safety, and inaccurate judgment existing in the prior art, and provides a strong guarantee for the safe and stable operation of the zero-sequence current transformer.

[0067] In an optional embodiment, training the state judgment model specifically includes:

[0068] Collect the historical deviation dataset \(V =\{(d 1 ,y 1 ),(d 2 ,y 2 ),...,(d n ,y n )\}\), where \(d n \) is the \(n\)th historical deviation feature vector, and \(y n \in\{Normal, Aging Warning, Fault\}\) is the status label;

[0069] Build a normal state SVM classifier and an aging warning SVM classifier. With normal state samples as the positive class and aging warning and fault state samples as the negative class, train the normal state SVM classifier through the optimization objective of maximizing the classification margin to obtain the hyperplane \(H 1 \); with aging warning samples as the positive class and fault state samples as the negative class, train the aging warning SVM classifier through the optimization objective of maximizing the classification margin to obtain the hyperplane \(H 2 \).

[0070] Specifically, when training the normal state SVM classifier, assign the corresponding class label to each sample data. Set the normal state as the positive class (+1), and unify the aging warning and fault state samples as the negative class (-1). Its optimization training objective is to find a hyperplane \(H 1 \) that can effectively separate the normal state (positive class) samples from the aging warning and fault state (negative class) samples. This hyperplane \(H 1 \) is represented by the following formula:

[0071] f 1 = w 1 * x + b 1

[0072] where \(w 1 \) is the normal vector of the hyperplane \(H 1 \), and \(b 1 \) is the bias term of \(H 1 \).

[0073] Similarly, with aging warning samples as the positive class and fault state samples as the negative class, train the hyperplane \(H 2 \). This hyperplane \(H 2 \) is represented by the following formula:

[0074] f 2 = w 2 * x + b 2

[0075] where \(w 2 \) is the normal vector of the hyperplane \(H 2 \), and \(b 2 \) is the bias term of \(H 2The paranoid term.

[0076] The hyperplane H is achieved by solving the following optimization problem 1 and the hyperplane H 2 training:

[0077]

[0078] y i (w i *x i +b i )≥1

[0079] Solve by the Lagrange multiplier method or the sequential minimal optimization algorithm to obtain the corresponding w 1 、w 2 、b 1 、b 2 。

[0080] In an optional embodiment, the deviation feature vector is input into a pre-trained state judgment model to output the state judgment result of the zero-sequence current transformer, specifically including: inputting the deviation feature vector D into a normal state SVM classifier to obtain a first result value. If the first result value is less than a first threshold, it indicates that the zero-sequence current transformer is in a normal state; if the first result value is greater than the first threshold, the deviation feature vector D is input into an aging warning SVM classifier to obtain a second result value. If the second result value is less than a second threshold, it indicates that the zero-sequence current transformer is in an aging state; if the second result value is greater than the second threshold, it indicates that the zero-sequence current transformer is in a faulty state.

[0081] When judging the state of the zero-sequence current transformer, inputting the deviation feature vector into the pre-trained state judgment model is the core step to achieve accurate diagnosis. Specifically, first input the deviation feature vector into the normal state SVM classifier. Through the calculation of this classifier, a first result value can be obtained. The magnitude of this result value directly reflects the degree of proximity of the current state of the zero-sequence current transformer to the normal state. If the first result value is less than a pre-set first threshold, it indicates that the zero-sequence current transformer is in good operating condition and in a normal working state, and can continue to operate stably without further inspection or maintenance.

[0082] However, if the first result value is greater than the first threshold, this implies that there may be potential problems with the zero-sequence current transformer and further diagnosis is required. At this time, the same deviation feature vector is input into the aging warning SVM classifier, and through the analysis of this classifier, a second result value can be obtained. Similarly, the magnitude of the second result value reflects the degree of association between the current state of the zero-sequence current transformer and the aging warning state. If the second result value is less than the preset second threshold, this indicates that although the zero-sequence current transformer is not in a normal state, the problem is not serious and it is in the aging warning state. In this case, regular monitoring and maintenance are recommended to prevent the problem from deteriorating further.

[0083] Conversely, if the second result value is greater than the second threshold, this indicates that the zero-sequence current transformer has developed a relatively serious fault and immediate repair or replacement is required. Through this hierarchical judgment method, the state of the zero-sequence current transformer can be effectively and accurately classified to ensure the safe and stable operation of the equipment. This process not only improves the accuracy of state judgment but also provides a scientific basis for subsequent maintenance and management, effectively avoiding potential risks and economic losses caused by equipment failures.

[0084] Furthermore, the first threshold and the second threshold are dynamic values. The setting of the first threshold and the second threshold specifically includes: substituting the historical deviation feature vector d n into the decision function f 1 of the hyperplane H 1 (d n ) and the decision function f 2 of the hyperplane H 2 (d n ) to obtain the first decision value and the second decision value respectively. Aggregate multiple first decision values into a first array and multiple second decision values into a second array. Calculate the 99% quantile of the first array as the first threshold and calculate the 50% quantile of the second array as the second threshold.

[0085] In an optional implementation manner, in step S4, corresponding control operations are performed according to the state judgment result, specifically including:

[0086] When the zero-sequence current transformer is in a normal state, generate a self-check qualified report;

[0087] When the zero-sequence current transformer is in an aging state, generate a marked transformer identifier;

[0088] When the zero-sequence current transformer is in a fault state, trigger a locking signal to disconnect the output circuit of the zero-sequence current transformer

[0089] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. The protection scope of the present invention shall be subject to the protection scope of the said claims.

Claims

1. A method for determining the state of a zero-sequence transformer, characterized in that: The method comprises the following steps: According to a preset cycle or an external instruction, real-time operating parameters of the zero-sequence transformer are obtained, wherein the operating parameters at least include zero-sequence current, zero-sequence voltage, ambient temperature and impedance parameters of the transformer winding; Preprocessing the operating parameters, and comparing the preprocessing results with pre-stored reference parameters to generate a deviation feature vector; Inputting the deviation feature vector into a pre-trained state judgment model, and outputting a state judgment result of the zero-sequence mutual inductor, wherein the state judgment result includes a normal state, a warning state, and a fault state; Execute corresponding control operations according to the state judgment result.

2. The method for determining the state of a zero-sequence transformer according to claim 1, characterized in that: According to a preset cycle or external instructions, the real-time operating parameters of the zero-sequence transformer are obtained, specifically including: according to the preset cycle or external instructions, a fixed-size simulated leakage signal is generated for the zero-sequence coil, and based on the simulated leakage signal, the real-time operating parameters of the zero-sequence transformer are obtained.

3. The method for determining the state of a zero-sequence transformer according to claim 2, characterized in that: Preprocessing the operating parameters and comparing the preprocessing results with pre-stored reference parameters to generate a deviation feature vector specifically includes: Performing data cleaning on the operating parameters to remove abnormal values ​​and noise data; Normalize the cleaned data to obtain real-time operating parameters; The real-time operating parameters are compared with pre-stored reference parameters to generate a deviation feature vector.

4. The method for determining the state of a zero-sequence transformer according to claim 3, characterized in that: Comparing the real-time operating parameters with the pre-stored reference parameters to generate a deviation feature vector specifically includes: During the factory calibration of the zero-sequence transformer, the zero-sequence current I ref , zero sequence voltage U ref , zero sequence impedance Z ref ; The current relative deviation ΔI, voltage relative deviation ΔU, and impedance relative deviation ΔZ are calculated by the following formulas: Among them I A is the real-time leakage current, U A is the real-time leakage voltage, Z A is the real-time impedance; The current relative deviation ΔI, voltage relative deviation ΔU, impedance relative deviation ΔZ, and ambient temperature are combined to form a deviation feature vector D = {ΔI i ,ΔU i ,ΔZ i ,T}.

5. The method for determining the state of a zero-sequence transformer according to claim 4, characterized in that: Training the state judgment model includes: Collect the historical deviation data set V = {(d1,y1),(d2,y2),...,(d n ,y n )}, where d n is the nth historical deviation feature vector, y n ∈{normal, aging warning, fault} is the state label; A normal state SVM classifier and an aging warning SVM classifier were built. With normal state samples as the positive class and aging warning and fault state samples as the negative class, the normal state SVM classifier was trained by maximizing the optimization goal of the classification interval to obtain the hyperplane H1; with aging warning samples as the positive class and fault state samples as the negative class, the aging warning SVM classifier was trained by maximizing the optimization goal of the classification interval to obtain the hyperplane H2.

6. The method for determining the state of a zero-sequence transformer according to claim 5, characterized in that: The deviation feature vector is input into a pre-trained state judgment model, and the state judgment result of the zero-sequence transformer is output, which specifically includes: inputting the deviation feature vector D into a normal state SVM classifier to obtain a first result value. If the first result value is less than a first threshold value, it indicates that the zero-sequence transformer is in a normal state; if the first result value is greater than the first threshold value, the deviation feature vector D is input into an aging warning SVM classifier to obtain a second result value. If the second result value is less than the second threshold value, it indicates that the zero-sequence transformer is in an aging state. If the second result value is greater than the second threshold value, it indicates that the zero-sequence transformer is in a fault state.

7. The method for determining the state of a zero-sequence transformer according to claim 6, characterized in that: Setting the first threshold and the second threshold specifically includes: n Substitute the decision function d1(d n ) and the decision function f2(d n ), respectively obtain a first decision value and a second decision value, aggregate multiple first decision values ​​into a first array, aggregate multiple second decision values ​​into a second array, calculate the 99% quantile of the first array as the first threshold, and calculate the 50% quantile of the second array as the second threshold.

8. The method for determining the state of a zero-sequence transformer according to claim 7, characterized in that: When the zero-sequence transformer is in normal state, a self-test qualified report is generated; When the zero-sequence transformer is in an aging state, a transformer identification mark is generated; When the zero-sequence transformer is in a fault state, a blocking signal is triggered to disconnect the output circuit of the zero-sequence transformer.

Images