High-voltage circuit breaker fault diagnosis method and system and computer equipment

Abstract

The invention relates to the technical field of power systems, and discloses a high-voltage circuit breaker fault diagnosis method and system and computer equipment, and the method comprises the steps: analyzing the change waveform of contact displacement and coil current during the operation of a high-voltage circuit breaker; constructing a high-voltage circuit breaker fault data set covering four fault situations of coil voltage, coil resistance, a damper and a spring structure; building a preliminary SVM fault classification diagnosis model; fusing a data relationship between faults and a parallel acceleration module into the preliminary SVM fault classification diagnosis model, and training the preliminary SVM fault classification diagnosis model to obtain a trained SVM fault classification diagnosis model; and diagnosing the high-voltage circuit breaker according to the trained SVM fault classification diagnosis model to obtain a fault diagnosis result. By analyzing the change waveform of the contact displacement and the coil current during the operation of the high-voltage circuit breaker, the fault data set covering various fault conditions is constructed, and the SVM fault classification diagnosis model is adopted for diagnosis, so that various faults of the high-voltage circuit breaker can be accurately diagnosed.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a high-voltage circuit breaker fault diagnosis method, system and computer equipment. Background Art

[0002] High voltage circuit breakers (HVCBs) are essential electrical switching devices in power systems and play an important protective role in power grids. Their main function is to cut off or close the circuit. The reliable operation of these key components improves the safety of the power system. Therefore, ensuring the normal operation of HVCBs is the key to ensuring the safety of power systems. According to the reliability survey results, the operating structures such as control circuits, auxiliary circuits, and dampers are the main parts of HVCB mechanical failures. Proposing and implementing a comprehensive and intelligent method to detect HVCB failures is the main research work at present.

[0003] In the existing technology, the traditional HVCB mechanical state detection methods mainly include regular maintenance and post-fault maintenance. Regular maintenance is time-consuming and labor-intensive, and it is difficult to formulate reasonable maintenance intervals, and it may even damage the equipment. Post-fault maintenance may cause the entire equipment to be paralyzed, and even affect the normal operation of the entire power system, causing huge economic losses and serious social impacts. Artificial intelligence methods mainly include neural networks, fuzzy control, Bayesian algorithms, etc. A network model is constructed to analyze the various detection data of the HVCB and summarize the various faults that occur in the HVCB. However, these algorithms are prone to model overfitting when analyzing data with long time series. Therefore, it is necessary to optimize the algorithm and data processing.

[0004] Therefore, it is necessary to improve the existing technology.

[0005] The above information is presented as background information only to assist with understanding the present disclosure and no determination or admission is made as to whether any of the above may be used as prior art with respect to the present disclosure. Summary of the invention

[0006] The present invention provides a high-voltage circuit breaker fault diagnosis method, system and computer equipment to solve the problems existing in the prior art.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A high voltage circuit breaker fault diagnosis method comprises the following steps:

[0009] Analyze the changing waveforms of contact displacement and coil current during operation of high-voltage circuit breakers;

[0010] Construct a high-voltage circuit breaker fault dataset covering four fault scenarios: coil voltage, coil resistance, damper, and spring structure;

[0011] Build a preliminary SVM fault classification diagnosis model;

[0012] Integrate the data relationship between faults and the parallel acceleration module into the preliminary SVM fault classification diagnosis model, and train it to obtain the trained SVM fault classification diagnosis model;

[0013] The high-voltage circuit breaker is diagnosed based on the trained SVM fault classification diagnosis model to obtain the fault diagnosis results.

[0014] Optionally, the analyzing of the contact displacement and coil current waveform changes during operation of the high-voltage circuit breaker includes:

[0015] Use simulation software to simulate high-voltage circuit breaker signals and obtain the waveforms of coil current and contact displacement;

[0016] According to predetermined rules, record relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform;

[0017] The relevant parameters of the coil current waveform include changes in the relevant parameters of the coil current waveform when the disconnection command is received, specifically including:

[0018] The current value, rise rate and rise time when the current rises to the first maximum value in the first duration; the current value, fall rate and fall time when the current decreases to the minimum value after the second duration; the current value, rise rate and rise time when the current increases to the second maximum value in the third duration; and the current fall rate and de-energization time during the process of the auxiliary contact causing the coil de-energization current to drop to zero;

[0019] The relevant parameters of the contact displacement waveform include the changes of the relevant parameters of the contact displacement waveform when the command is received, specifically including:

[0020] The starting value and displacement rate of the displacement from the time the command is received to the time the latch in the release structure lasts; the displacement change rate and displacement change amount in the process when the spring starts to release energy and move the contacts; the separation displacement value and separation time of the arc contact separation when the displacement reaches the contact separation displacement; the final displacement value and total displacement duration until the high-voltage circuit breaker reaches its fully open state; the rebound displacement value, rebound rate, extension displacement value, extension rate and related parameters of the rebound displacement change due to the small rebound and extension displacement caused by the spring inertia during the spring rebound time, the related parameters of the rebound displacement change include the rebound amplitude and extension amplitude, etc.

[0021] Optionally, the constructing of a high-voltage circuit breaker fault data set covering four fault scenarios of coil voltage, coil resistance, damper and spring structure comprises:

[0022] By changing the coil resistance, coil voltage, damper and spring constant, four parameters of high-voltage circuit breaker are simulated to simulate fault conditions;

[0023] Obtain relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform under different fault conditions, and mark the relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform according to a predetermined marking rule: mark the four fault types as abnormal coil voltage, abnormal coil resistance, abnormal damper and abnormal spring structure, and mark the normal state as the normal state;

[0024] Creating multiple data sets, each data set includes samples corresponding to a normal state, a coil voltage abnormality, a coil resistance abnormality, a damper abnormality, and a spring structure abnormality;

[0025] Some samples are selected from each data set to form the training data set, and the remaining samples form the test data set.

[0026] Optionally, diagnosing the high-voltage circuit breaker according to the trained SVM fault classification diagnosis model includes:

[0027] Diagnosis based on hyperplane expressions;

[0028]

[0029] Where w is the weight vector; φ is the objective function; x i is the input of the sample; b is the optimal hyperplane offset;

[0030] Solve the expression of the hyperplane to obtain the optimal classification decision function of the sample;

[0031] Based on the SVM classifier, a maximum margin hyperplane between two types of boundary points is found in the training data set, where the maximum margin hyperplane is the maximum distance between the nearest data points in the two types of boundary points.

[0032] Optionally, the trained SVM fault classification diagnosis model includes an input layer, a dense layer, and an output layer, and Softplus is used as the activation function of the dense layer, and its expression is as follows:

[0033] ;

[0034] Among them, x is the variable input to the dense layer, and e is a natural constant.

[0035] Optionally, it also includes: improving the SVM fault classification diagnosis model;

[0036] The improvement to the SVM fault classification diagnosis model includes:

[0037] Compare and analyze the differences between abnormal coil voltage and abnormal coil resistance, calculate the ratio of the duration of the coil current waveform relative to the healthy state, and compare the variation range of the duration under different faults;

[0038] Compare and analyze the difference between damper failure and spring damage, and compare the extension time of the faulty spring with the contact separation displacement and healthy state for intuitive display;

[0039] Compare and analyze the differences between coil resistance failure and spring damage, take the maximum value of the relevant current as the decisive feature, and analyze its changes under different faults by drawing a ratio change diagram;

[0040] According to the difference analysis results between coil voltage abnormality and coil resistance abnormality, damper failure and spring damage, coil resistance failure and spring damage, after the first training of the SVM fault classification diagnosis model, the fault relationship is trained for the second time.

[0041] Optionally, it also includes: accelerating the processing of the SVM fault classification diagnosis model;

[0042] The accelerating process of the SVM fault classification diagnosis model includes:

[0043] To solve the problem of time-consuming training of SVM fault classification diagnosis model, the Laplace operator is used in combination with a parallel acceleration computing module consisting of one-dimensional feature fast transformation to improve the network.

[0044] Optionally, it also includes: verifying and evaluating the improved SVM fault classification diagnosis model;

[0045] The verification and evaluation of the improved SVM fault classification diagnosis model includes:

[0046] The training set data is input into the improved SVM fault classification diagnosis model for training. After the training is completed, a test data set containing a specific number of samples of each category is used for verification testing;

[0047] The fault diagnosis accuracy is evaluated by the classification results after diagnosis.

[0048] The present invention further provides a high-voltage circuit breaker fault diagnosis system, which is used to implement the high-voltage circuit breaker fault diagnosis method as described in any one of the above items, comprising:

[0049] Waveform analysis module, used to analyze the changing waveforms of contact displacement and coil current when the high-voltage circuit breaker is in operation;

[0050] A dataset construction module is used to construct a high-voltage circuit breaker fault dataset covering four fault scenarios: coil voltage, coil resistance, damper, and spring structure;

[0051] Model building module, used to build a preliminary SVM fault classification diagnosis model;

[0052] A model training module is used to integrate the data relationship between faults and the parallel acceleration module into the preliminary SVM fault classification diagnosis model, and train it to obtain the trained SVM fault classification diagnosis model;

[0053] The fault diagnosis module is used to diagnose the high-voltage circuit breaker based on the trained SVM fault classification diagnosis model to obtain the fault diagnosis result.

[0054] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the high-voltage circuit breaker fault diagnosis method as described in any one of the above items when executing the computer program.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention provides a high-voltage circuit breaker fault diagnosis method, system and computer equipment, which construct a fault data set covering various fault situations by analyzing the changing waveforms of contact displacement and coil current when the high-voltage circuit breaker is running, and use the SVM fault classification diagnosis model for diagnosis, so as to accurately diagnose various faults of the high-voltage circuit breaker.

[0057] The present invention has other features and advantages, which will be apparent from or will be described in detail in the accompanying drawings and the following detailed description incorporated herein, which together serve to explain the specific principles of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 It is a flow chart of a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0060] Figure 2 It is a logic diagram of a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0061] Figure 3 It is a flow chart of step S1 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0062] Figure 4 It is a waveform diagram of coil current CC in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0063] Figure 5 It is a waveform diagram of contact displacement CT in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0064] Figure 6 It is a flow chart of step S2 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0065] Figure 7 It is a CC and CT waveform change diagram under four fault states in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0066] Figure 8 It is a flow chart of step S5 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0067] Fig. 9 It is a ratio diagram of TP1 of F1 and F2 relative to F0 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0068] Fig.10 It is a ratio diagram of Over-Tr of F3 and F4 relative to F0 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0069] Fig.11 It is a ratio diagram of T-damp of F3 and F4 relative to F0 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0070] Fig.12 It is a ratio diagram of IP1 of F2 and F4 relative to F0 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0071] Fig.13 It is a flow chart of step S6 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0072] Fig.14 It is a flow chart of step S7 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0073] Fig.15 It is a flow chart of step S8 in a high-voltage circuit breaker fault diagnosis method provided by the present invention;

[0074] Fig.16 The present invention provides a structural block diagram of a high-voltage circuit breaker fault diagnosis system.

[0075] Figure numerals: 10, waveform analysis module; 20, data set construction module; 30, model building module; 40, model training module; 50, fault diagnosis module. DETAILED DESCRIPTION

[0076] In order to explain in detail the possible application scenarios, technical principles, specific schemes that can be implemented, and the purposes and effects that can be achieved, the following is a detailed description of the specific embodiments listed in conjunction with the accompanying drawings. The embodiments described herein are only used to more clearly illustrate the technical solutions of the present application, and are therefore only used as examples, and cannot be used to limit the scope of protection of the present application.

[0077] Reference to "embodiment" herein means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present application. The term "embodiment" appearing in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or association with other embodiments. In principle, in the present application, as long as there is no technical contradiction or conflict, the various technical features mentioned in the embodiments can be combined in any way to form a corresponding implementable technical solution.

[0078] Unless otherwise defined, the technical terms used in this document have the same meanings as those generally understood by those skilled in the art to which this application belongs; the use of relevant terms in this document is only for describing specific embodiments and is not intended to limit this application.

[0079] In the description of this application, the term "and / or" is an expression used to describe the logical relationship between objects, indicating that three relationships may exist, for example, A and / or B, which means: A exists, B exists, and A and B exist at the same time. In addition, the character " / " in this article generally indicates that the objects before and after are in an "or" logical relationship.

[0080] In the present application, terms such as “first” and “second” are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship of quantity, priority or sequence between these entities or operations.

[0081] Without further limitations, in this application, the words "include", "comprises", "has" or other similar expressions used in the sentences are intended to cover non-exclusive inclusion. These expressions do not exclude the presence of additional elements in the process, method or product including the elements, so that the process, method or product including a series of elements may include not only those limited elements, but also other elements not explicitly listed, or also include elements inherent to such process, method or product.

[0082] Similar to the understanding in the Examination Guidelines, in this application, expressions such as "greater than", "less than", "exceed" and the like are understood to exclude the number itself; expressions such as "above", "below", "within" and the like are understood to include the number itself. In addition, in the description of the embodiments of this application, "multiple" means more than two (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups", "multiple times", etc., unless otherwise clearly and specifically limited.

[0083] In the description of the embodiments of the present application, space-related expressions used, such as "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "vertical", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicate the orientation or position relationship based on the orientation or position relationship shown in the specific embodiments or drawings, and are only for the convenience of describing the specific embodiments of the present application or facilitating the reader's understanding, and do not indicate or imply that the referred device or component must have a specific position, a specific orientation, or be constructed or operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present application.

[0084] Unless otherwise expressly specified or limited, in the description of the embodiments of the present application, the terms such as "install", "connect", "connect", "fix", "set", etc. used should be understood in a broad sense. For example, the "connection" can be a fixed connection, a detachable connection, or an integrated setting; it can be a mechanical connection, an electrical connection, or a communication connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium; it can be the internal connection of two elements or the interaction relationship between two elements. For technicians in the technical field to which the present application belongs, the specific meanings of the above terms in the embodiments of the present application can be understood according to the specific circumstances.

[0085] The purpose of the present invention is to provide a method for high-voltage circuit breaker (HVCB) fault diagnosis based on an improved support vector machine (SVM). It can solve the problem that the existing model is not accurate and the model is prone to overfitting when analyzing a large amount of data, and utilizes the waveforms of the coil current (CC) and contact displacement (CT) of the HVCB during operation. Data under four fault conditions, namely, coil voltage, coil resistance, damper and spring structure faults, are collected, and the SVM fault classification diagnosis model is used for training and testing. At the same time, the model is improved by adding the relationship between several fault data and a parallel acceleration module to improve the accuracy of the model fault diagnosis and the training speed. It overcomes the problem that the model in the prior art is prone to overfitting and the prediction accuracy is not high.

[0086] Please refer to Figure 1 , Figure 2The embodiment of the present invention provides a high-voltage circuit breaker fault diagnosis method, comprising the following steps:

[0087] S1. Analyze the changing waveforms of contact displacement and coil current during operation of high-voltage circuit breaker.

[0088] S2. Construct a high-voltage circuit breaker fault dataset covering four fault scenarios: coil voltage, coil resistance, damper, and spring structure.

[0089] S3. Build a preliminary SVM fault classification diagnosis model.

[0090] S4. Integrate the data relationship between faults and the parallel acceleration module into the preliminary SVM fault classification diagnosis model, and train it to obtain the trained SVM fault classification diagnosis model.

[0091] S5. Diagnose the high-voltage circuit breaker based on the trained SVM fault classification diagnosis model to obtain a fault diagnosis result.

[0092] The training set data was input into the improved SVM model for training, and the test data set containing a specific number of samples for each category was used for validation testing. The fault diagnosis accuracy was evaluated by the classification results after diagnosis. The improved SVM fault diagnosis classifier had a diagnosis accuracy of more than 90% for the four faults of coil voltage, coil resistance, damper and spring structure, indicating that the improved support vector machine model can better diagnose the fault type of HVCB, thereby accurately diagnosing the fault type of high-voltage circuit breaker, providing strong support for maintaining the safety and stability of the power system.

[0093] It can be understood that in this embodiment, the waveforms of the coil current (CC) and contact displacement (CT) of the HVCB during operation are used to collect data under four fault conditions, namely, coil voltage, coil resistance, damper and spring structure faults, and the SVM fault classification diagnosis model is used for training and testing. At the same time, the model is improved by adding the relationship between several fault data, and comparing it with the data in the healthy state, so as to improve the accuracy of the model in diagnosing faults.

[0094] Please refer to Figure 3 , wherein step S1 analyzes the changing waveforms of contact displacement and coil current when the high-voltage circuit breaker is running, including:

[0095] S11, using simulation software to simulate the high-voltage circuit breaker signal to obtain the waveforms of the coil current and the contact displacement;

[0096] S12. Record relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform according to a predetermined rule.

[0097] In step S1, a 10kV sulfur hexafluoride circuit breaker (SF6 HVCB) is used as an example for research and testing. Since the changes in contact displacement (CT) and coil current (CC) during the operation of the HVCB can directly reflect its working state, the waveform changes of CT and CC are taken as the research objects of HVCB fault diagnosis.

[0098] Specifically, in this step, the waveform is obtained by using simulation software: the HVCB signal is simulated using Matlab Simulink software to obtain the waveforms of the coil current CC and the contact displacement CT.

[0099] The relevant parameters of the coil current waveform include changes in relevant parameters of the coil current waveform when the disconnection command is received, specifically including:

[0100] The current value, rise rate and rise time when the current rises to the first maximum value in the first duration; the current value, fall rate and fall time when the current decreases to the minimum value after the second duration; the current value, rise rate and rise time when the current increases to the second maximum value in the third duration; and the current fall rate and de-energization time during the process of the auxiliary contact causing the coil de-energization current to drop to zero;

[0101] The relevant parameters of the contact displacement waveform include the changes of the relevant parameters of the contact displacement waveform when the command is received, specifically including:

[0102] The starting value and displacement rate of the displacement from the time the command is received to the time the latch in the release structure lasts; the displacement change rate and displacement change amount in the process when the spring starts to release energy and move the contacts; the separation displacement value and separation time of the arc contact separation when the displacement reaches the contact separation displacement; the final displacement value and total displacement duration until the high-voltage circuit breaker reaches its fully open state; the rebound displacement value, rebound rate, extension displacement value, extension rate and related parameters of the rebound displacement change due to the small rebound and extension displacement caused by the spring inertia during the spring rebound time, the related parameters of the rebound displacement change include the rebound amplitude and extension amplitude, etc.

[0103] For the coil current waveform, when the disconnection command is received, the coil current increases to the first maximum value IP1 within the first duration TP1; then, the current decreases to its minimum value after the duration TV; subsequently, in the duration TP2, the current increases to the second maximum value IP2; finally, the auxiliary contact de-energizes the coil and the current drops to zero. Record the current value, rise rate, and rise time when the current rises to the first maximum value in the first duration; the current value, fall rate, and fall time when the current decreases to the minimum value after the second duration; the current value, rise rate, and rise time when the current increases to the second maximum value in the third duration; and the current fall rate and de-energization time in the process of the auxiliary contact de-energizing the coil current to zero.

[0104] Please refer to Figure 4 and Figure 5 , are the waveforms of coil current CC and contact displacement CT, respectively. Among them, for the contact displacement waveform, T-React refers to the duration from receiving the command to releasing the latch in the structure. During the duration, the spring begins to release energy and move the contacts. When the displacement reaches Sep-Tr, the arc contacts separate until the HVCB reaches its fully open state. During the T-daMP24025381 time, there will be a small rebound due to the inertia of the spring, resulting in displacement changes of Over-Tr and Re-Tr. Record the initial displacement value and displacement rate during the duration from receiving the command to releasing the latch in the structure; the displacement change rate and displacement change amount during the process when the spring starts to release energy and move the contacts; the separation displacement value and separation time of the arc contact separation when the displacement reaches the contact separation displacement; the final displacement value and total displacement duration until the high-voltage circuit breaker reaches its fully open state; the rebound displacement value, rebound rate, extension displacement value, extension rate and related parameters of the rebound displacement change, such as the rebound amplitude and extension amplitude, etc., due to the small rebound and extension displacement caused by the spring inertia during the spring rebound time.

[0105] Please refer to Figure 6 , further, in step S2, a high-voltage circuit breaker fault data set covering four fault conditions of coil voltage, coil resistance, damper and spring structure is constructed, including:

[0106] S21, simulating fault conditions by changing four parameter indicators of the high-voltage circuit breaker, namely, coil resistance, coil voltage, damper and spring constant;

[0107] S22, obtaining relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform under different fault conditions, and marking the relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform according to a predetermined marking rule;

[0108] Specifically, the marking method is as follows: the four fault types are marked as abnormal coil voltage, abnormal coil resistance, abnormal damper and abnormal spring structure, and the normal state is marked as normal state;

[0109] S23, creating multiple data sets, each data set including samples corresponding to a normal state, a coil voltage abnormality, a coil resistance abnormality, a damper abnormality, and a spring structure abnormality;

[0110] S24. Select some samples from each data set to form a training data set, and the remaining samples to form a test data set.

[0111] Among them, step S2 establishes a data set about HVCB faults, increases and decreases the four important parameter indicators of HVCB, namely coil resistance, coil voltage, damper and spring constant, simulates these four fault conditions, and realizes fault analysis of HVCB. And obtains the waveforms of CT and CC in these conditions, such as Figure 7 As shown, it can be seen that when relevant parameters change, such as the coil resistance increases, the coil voltage decreases, and the spring constant changes, the waveforms of CT and CC change significantly compared to normal conditions.

[0112] The test can diagnose the HVCB fault by analyzing the difference between CC and CT data under fault state and actual data, and identify the specific fault type.

[0113] The four fault types are labeled as F1 (coil voltage abnormality), F2 (coil resistance abnormality), F3 (damper abnormality) and F4 (spring structure abnormality), and the normal state is labeled as F0 for comparison. F1 is generated by random variation of coil voltage of simulated healthy HVCB. The coil voltage deviation range of generating F1 dataset is between 0.4V nominal − 1.7V nominal. Subsequently, F2 is created by random variation of coil resistance in the range of 0.5R nominal − 1.55 nominal of the healthy dataset. F3 and F4 are made by the same procedure. The deviation range of damper coefficient is between 0.3d nominal − 2d nominal. The variation range of spring constant is between 0.5 nominal and 2.1 nominal.

[0114] Five data sets were created, each including 135 samples, namely F0, F1, F2, F3 and F4, with a total of 675 sample data. The healthy and common failure modes of HVCB are processed. The data indicators in the CC and CT waveforms of HVCB are used to diagnose the specific fault type or normal state. The specific data indicators and data ranges are shown in Table 1.

[0115] Table 1 Diagnostic indicators and ranges

[0116]

[0117] The training datasets selected from the F0, F1, F2, F3, and F4 datasets consist of 95 samples each. The test dataset includes the remaining 40 samples from each dataset. Therefore, 475 samples are used for training and validation databases respectively, and 200 samples are used to test the discriminant algorithm. Table 2 shows the sample labels of the training and test datasets.

[0118] Table 2 Sample labels of training and test datasets

[0119]

[0120] Please refer to Figure 8 In step S5, the high-voltage circuit breaker is diagnosed according to the trained SVM fault classification diagnosis model, including:

[0121] S51, diagnosis based on hyperplane expression;

[0122] The expression of the hyperplane is as follows: ;

[0123] Where w is the weight vector; φ is the objective function; x i is the input of the sample; b is the optimal hyperplane offset;

[0124] S52, solving the expression of the hyperplane to obtain the optimal classification decision function of the sample;

[0125] S53. Find a maximum margin hyperplane between two types of boundary points in the training data set based on the SVM classifier, where the maximum margin hyperplane is the maximum distance between the nearest data points in the two types of boundary points.

[0126] Among them, support vector machine (SVM) is a machine learning method. Its basic idea is to solve the optimal hyperplane that meets the conditions in the high-dimensional feature space, which can accurately divide the training samples. After solving the expression of the hyperplane, the optimal classification decision function of the sample can be obtained. The SVM classifier finds the "maximum margin" hyperplane between the boundary points of two categories in the training data set. The maximum margin refers to the maximum distance between the closest data points in the two categories.

[0127] One of the advantages of using neural networks is that they classify data sets by finding failure patterns in them without requiring extensive knowledge of the system.

[0128] Specifically, the trained SVM fault classification diagnosis model includes an input layer, a dense layer, and an output layer. Softplus is used as the activation function of the dense layer, and its expression is as follows:

[0129] ;

[0130] Among them, x is the variable input to the dense layer, and e is a natural constant.

[0131] The SVM model takes the training set data as input and outputs four fault types of HVCB, namely: F0, F1, F2, F3 and F4. First, the parameters of the support vector machine fault diagnosis model are determined. The weight vectors w and b are optimized through cross validation. The optimization range of the parameters is set to [-10 10] and the iteration step is 0.5.

[0132] At the same time, the SVM classifier is improved. Based on the existing data, the relationship between several fault categories is increased to improve the diagnosis results.

[0133] Further analysis and comparison are carried out below.

[0134] A. Difference between coil voltage (F1) and coil resistance (F2):

[0135] Comparing the defective features with the healthy features, F1 increases the TP1 and TP2 of the CC waveform, while reducing (increasing) the three second features (Dt, IP 1, and IP 2), while the changes in TP1 and TP2 of the CC waveform in the F2 case are smaller. In order to compare the range of changes in TP1 in the training samples under F1 and F2, the ratio of the faulty TP1 to the healthy state is calculated, as Fig. 9 shown.

[0136] Depend on Fig. 9 It can be seen that the 95 samples belonging to F1 (red track) show greater changes compared to the other 95 samples in F2 (blue track). The ratio of F2 to the healthy state basically remains around 1, while the ratio in the F1 case is around 0.8 and 1.3.

[0137] B. Difference between damper failure (F3) and spring damage (F4):

[0138] Similarly, comparing the defective features with the healthy features, the extension time Over-Tr of the F3 and F4 fault springs is quite different from the displacement of contact separation. Ratioing it with the healthy state F0 can be more intuitively presented, as shown in the following example: Fig.10 and Fig.11 shown.

[0139] Depend on Fig.10 It can be seen that the 95 samples belonging to F4 (red track) show greater changes compared to the other 95 samples in F3 (blue track). The ratio of F4 to the healthy state is basically maintained at around 1, while the ratio in the case of F3 is around 0.5 and 3.5.

[0140] Depend on Fig.11 It can be seen that the 95 samples belonging to F1 (red track) show greater changes compared to the other 95 samples in F3 (blue track). The ratio of F4 relative to the healthy state is basically maintained at around 1, while the ratio in the case of F3 is around 0.9 and 1.4.

[0141] C. Difference between coil resistance failure (F2) and spring damage (F4):

[0142] Comparing the defective signature with the healthy one, IP1 or IP2 can be considered as the decisive signature, in case of F3 or F4, IP1 is almost constant, while in fault conditions associated with F1 or F2, it changes. Fig.12 Represents the change in ratio of IP1, F2 (blue) and F4 (red) under fault conditions to their normal values.

[0143] like Fig.12 As shown, the 95 samples belonging to F4 (red track) show greater changes compared to the other 95 samples in F2 (blue track). The ratio of F4 relative to the healthy state remains basically around 1, while the ratio in the case of F2 is around 0.8 and 1.5.

[0144] According to the analysis results of A, B, and C, the relationship between the above three faults is added to the SVM model. After the first training of the training set, the fault relationship is trained for the second time, so that several fault types can be further diagnosed and classified during the test verification process, thus improving the diagnostic performance of the entire model.

[0145] In order to solve the problem of time-consuming SVM model training, the Laplace operator is used in combination with the GPU parallel acceleration computing module to improve the network. This module consists of a one-dimensional feature fast Fourier transform. Considering the differential properties of the Fourier transform, the Laplace operator of the function u(x, y) can be calculated in the Fourier domain:

[0146]

[0147] in represents the continuous Fourier transform of u(x, y), and and is the continuous frequency.

[0148] The parallel development environment consists of the high-performance math library HPML, the parallel program support library, etc. It uses the OpenMP24025381 runtime library to optimize dynamic scheduling and load balancing, thread nesting parallelism, and multi-mode thread management. Adding a parallel acceleration module to the SVM model can significantly reduce the time it takes to train the SVM model.

[0149] The training set data is input into the improved SVM model for training. The test data set includes 40 samples of each category. After the training is completed successfully, a verification test is performed on it. The classification results after diagnosis are shown in Table 3.

[0150] Table 3 Test classification results

[0151]

[0152] It can be seen from Table 3 that the improved SVM fault diagnosis classifier has an accuracy of more than 90% for the four faults of coil voltage, coil resistance, damper and spring structure. Therefore, the improved support vector machine model can better diagnose the fault type of HVCB.

[0153] Please refer to Fig.13 Furthermore, the high-voltage circuit breaker fault diagnosis method in this embodiment also includes:

[0154] S6. Improvement on SVM fault classification diagnosis model.

[0155] Specifically, in step S6, the SVM fault classification diagnosis model is improved, including:

[0156] Compare and analyze the differences between abnormal coil voltage and abnormal coil resistance, calculate the ratio of the duration of the coil current waveform relative to the healthy state, and compare the variation range of the duration under different faults;

[0157] Compare and analyze the difference between damper failure and spring damage, and compare the extension time of the faulty spring with the contact separation displacement and healthy state for intuitive display;

[0158] Compare and analyze the differences between coil resistance failure and spring damage, take the maximum value of the relevant current as the decisive feature, and analyze its changes under different faults by drawing a ratio change diagram;

[0159] According to the difference analysis results between coil voltage abnormality and coil resistance abnormality, damper failure and spring damage, coil resistance failure and spring damage, after the first training of the SVM fault classification diagnosis model, the fault relationship is trained for the second time.

[0160] Please refer to Fig.14 Furthermore, the high-voltage circuit breaker fault diagnosis method in this embodiment also includes:

[0161] S7. Accelerate the processing of the SVM fault classification diagnosis model.

[0162] Specifically, in step S7, the SVM fault classification diagnosis model is accelerated, including:

[0163] To solve the problem of time-consuming training of SVM fault classification diagnosis model, the Laplace operator is used in combination with a parallel acceleration computing module composed of one-dimensional feature fast transformation to improve the network processing.

[0164] Among them, in order to solve the problem that the training of SVM model is time-consuming, this embodiment uses Laplace operator and GPU parallel acceleration computing module to improve the network. This module is composed of one-dimensional feature fast Fourier transform. Considering the differential properties of Fourier transform, the Laplace operator of function u (x, y) can be calculated in Fourier domain. The parallel development environment includes high-performance mathematics library HPML, parallel program support library, etc., and uses OpenMP24025381 runtime library for calculation. It optimizes dynamic scheduling and load balancing, thread nested parallelism, multi-mode thread management, etc. to reduce the time of SVM model in the training process.

[0165] Please refer to Fig.15 Furthermore, the high-voltage circuit breaker fault diagnosis method in this embodiment also includes:

[0166] S8. Verify and evaluate the improved SVM fault classification diagnosis model.

[0167] Specifically, in step S8, the improved SVM fault classification diagnosis model is verified and evaluated, including:

[0168] The training set data is input into the improved SVM fault classification diagnosis model for training. After the training is completed, a test data set containing a specific number of samples of each category is used for verification testing;

[0169] The fault diagnosis accuracy is evaluated by the classification results after diagnosis.

[0170] Please refer to Fig.16 Based on the above embodiments, the present invention further provides a high-voltage circuit breaker fault diagnosis system, which is used to implement any of the above high-voltage circuit breaker fault diagnosis methods, including:

[0171] The waveform analysis module 10 is used to analyze the changing waveforms of contact displacement and coil current when the high-voltage circuit breaker is in operation.

[0172] Specifically, the waveform analysis module 10 analyzes the waveform of the contact displacement and the coil current when the high-voltage circuit breaker is running, and uses professional simulation software to simulate the high-voltage circuit breaker signal to obtain the waveform of the coil current and the contact displacement. For the coil current waveform, the current value, rising rate, falling rate, rising time, falling time and other parameters at different time stages from the time the disconnection command is received are recorded, as well as the current falling rate and power-off time in the process of the auxiliary contact causing the coil power-off current to drop to zero. For the contact displacement waveform, the displacement starting value, displacement rate, displacement change rate, displacement change amount, separation displacement value, separation time, final displacement value, total displacement duration, rebound displacement value, rebound rate, extension displacement value, extension rate and related parameters of the rebound displacement change (such as rebound amplitude, extension amplitude, etc.) from the time the command is received to the different stages are recorded, thereby providing basic data for subsequent fault diagnosis. By analyzing these detailed parameters, the operating status of the high-voltage circuit breaker can be judged more accurately, providing a strong basis for subsequent fault diagnosis.

[0173] The data set construction module 20 is used to construct a high-voltage circuit breaker fault data set covering four fault scenarios: coil voltage, coil resistance, damper and spring structure.

[0174] The data set construction module 20 simulates fault conditions by changing the four important parameter indicators of the high-voltage circuit breaker, namely, coil resistance, coil voltage, damper and spring constant. The relevant parameters of the coil current waveform and the contact displacement waveform under different fault conditions are obtained, and according to the predetermined marking rules, the four fault types are marked as abnormal coil voltage, abnormal coil resistance, abnormal damper and abnormal spring structure, and the normal state is marked as normal state. Create multiple data sets, each of which includes samples corresponding to the normal state, abnormal coil voltage, abnormal coil resistance, abnormal damper and abnormal spring structure. Select some samples from each data set to form a training data set, and the remaining samples form a test data set.

[0175] The data set construction module 20 provides abundant fault data samples, so that the SVM fault classification diagnosis model can learn the characteristics of different fault types and improve the generalization ability and diagnosis accuracy of the model.

[0176] The model building module 30 is used to build a preliminary SVM fault classification diagnosis model.

[0177] The model building module 30 constructs a model structure including an input layer, a dense layer, and an output layer based on the machine learning method of the support vector machine (SVM). A specific function (such as Softplus) is used as the activation function of the dense layer. In the expression of the hyperplane, the weight vector w, the objective function φ, the sample input xi, and the optimal hyperplane offset b are clearly defined. The SVM classifier finds the "maximum margin" hyperplane between two types of boundary points in the training data set, that is, the maximum distance between the nearest data points in the two types of boundary points, which is conducive to providing an effective classification tool for fault diagnosis. By solving the optimal hyperplane in the high-dimensional feature space, samples of different fault types can be accurately divided, thereby improving the accuracy of fault diagnosis.

[0178] The model training module 40 is used to integrate the data relationship between faults and the parallel acceleration module into the preliminary SVM fault classification diagnosis model, and train it to obtain the trained SVM fault classification diagnosis model.

[0179] In the model training module 40, the differences between different fault types are compared and analyzed, such as abnormal coil voltage and abnormal coil resistance, damper fault and spring damage, coil resistance fault and spring damage. By calculating the ratio of related parameters, drawing a ratio change graph, etc., the fault relationship is added to the SVM model, and a second training is performed after the first training to improve the model diagnosis performance.

[0180] In addition, in order to solve the problem that the SVM model training calculation is time-consuming, the model training module 40 uses the Laplace operator and combines the parallel acceleration calculation module composed of one-dimensional feature fast Fourier transform to improve the network. It can be understood that the parallel development environment includes high-performance mathematical libraries, parallel program support libraries, etc., uses the OpenMP24025381 runtime library for calculation, and optimizes dynamic scheduling and load balancing, thread nesting parallelism, multi-mode thread management, etc. to reduce the time of the SVM model in the training process.

[0181] By incorporating data relationships between faults, the model can better learn the differences between different fault types and improve diagnostic accuracy. At the same time, the parallel acceleration module greatly improves the training efficiency of the model, making it more suitable for practical application scenarios.

[0182] The fault diagnosis module 50 is used to diagnose the high-voltage circuit breaker according to the trained SVM fault classification diagnosis model to obtain a fault diagnosis result.

[0183] The fault diagnosis module 50 inputs the training set data into the improved SVM model for training. After the training is completed, a test data set containing a specific number of samples of each category is used for verification testing. The fault diagnosis accuracy is evaluated by the classification results after diagnosis. Thus, the fault type of the high-voltage circuit breaker can be accurately diagnosed, providing timely and effective information for maintaining the safety and stability of the power system, helping to take corresponding maintenance measures and reduce the losses caused by the fault.

[0184] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements any of the above high-voltage circuit breaker fault diagnosis methods when executing the computer program.

[0185] Finally, it should be noted that although the above embodiments have been described in the specification and drawings of this application, this does not limit the scope of patent protection of this application. All technical solutions generated by replacing or modifying equivalent structures or equivalent processes based on the essential concept of this application using the contents recorded in the specification and drawings of this application, as well as directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, are included in the scope of patent protection of this application.

Claims

1. A high voltage circuit breaker fault diagnosis method, characterized in that: The following steps are involved: Analyze the changing waveforms of contact displacement and coil current during operation of high-voltage circuit breakers; Construct a high-voltage circuit breaker fault dataset covering four fault scenarios: coil voltage, coil resistance, damper, and spring structure; Build a preliminary SVM fault classification diagnosis model; Integrate the data relationship between faults and the parallel acceleration module into the preliminary SVM fault classification diagnosis model, and train it to obtain the trained SVM fault classification diagnosis model; The high-voltage circuit breaker is diagnosed based on the trained SVM fault classification diagnosis model to obtain the fault diagnosis results.

2. A high voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: The analysis of the change waveform of the contact displacement and the coil current during the operation of the high-voltage circuit breaker includes: Use simulation software to simulate high-voltage circuit breaker signals and obtain the waveforms of coil current and contact displacement; According to predetermined rules, record relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform; The relevant parameters of the coil current waveform include changes in the relevant parameters of the coil current waveform when the disconnection command is received, specifically including: The current value, rise rate and rise time when the current rises to the first maximum value in the first duration; the current value, fall rate and fall time when the current decreases to the minimum value after the second duration; the current value, rise rate and rise time when the current increases to the second maximum value in the third duration; and the current fall rate and de-energization time during the process of the auxiliary contact causing the coil de-energization current to drop to zero; The relevant parameters of the contact displacement waveform include the changes of the relevant parameters of the contact displacement waveform when the command is received, specifically including: The starting value and displacement rate of the displacement from the time the command is received to the time the latch in the release structure lasts; the displacement change rate and displacement change amount in the process when the spring starts to release energy and move the contacts; the separation displacement value and separation time of the arc contact separation when the displacement reaches the contact separation displacement; the final displacement value and total displacement duration until the high-voltage circuit breaker reaches its fully open state; the rebound displacement value, rebound rate, extension displacement value, extension rate and related parameters of the rebound displacement change due to the small rebound and extension displacement caused by the spring inertia during the spring rebound time, the related parameters of the rebound displacement change include the rebound amplitude and extension amplitude, etc.

3. A high voltage circuit breaker fault diagnosis method according to claim 2, characterized in that: The construction covers a high-voltage circuit breaker fault data set for four fault scenarios: coil voltage, coil resistance, damper, and spring structure, including: By changing the coil resistance, coil voltage, damper and spring constant, four parameters of high-voltage circuit breakers were simulated to simulate fault conditions; Obtain relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform under different fault conditions, and mark the relevant parameters of the coil current waveform and relevant parameters of the contact displacement waveform according to a predetermined marking rule: mark the four fault types as abnormal coil voltage, abnormal coil resistance, abnormal damper and abnormal spring structure, and mark the normal state as the normal state; Creating multiple data sets, each data set includes samples corresponding to a normal state, a coil voltage abnormality, a coil resistance abnormality, a damper abnormality, and a spring structure abnormality; Some samples are selected from each data set to form the training data set, and the remaining samples form the test data set.

4. The high voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: The high-voltage circuit breaker is diagnosed based on the trained SVM fault classification diagnosis model, including: Diagnosis based on hyperplane expressions; ； Where w is the weight vector; φ is the objective function; x i is the input of the sample; b is the optimal hyperplane offset; Solve the expression of the hyperplane to obtain the optimal classification decision function of the sample; Based on the SVM classifier, a maximum margin hyperplane between two types of boundary points is found in the training data set, where the maximum margin hyperplane is the maximum distance between the nearest data points in the two types of boundary points.

5. The high voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: The trained SVM fault classification diagnosis model includes an input layer, a dense layer, and an output layer. Softplus is used as the activation function of the dense layer, and its expression is as follows: ； Among them, x is the variable input to the dense layer, and e is a natural constant.

6. The high voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: Also includes: Improvement of SVM fault classification diagnosis model; The improvement to the SVM fault classification diagnosis model includes: Compare and analyze the differences between abnormal coil voltage and abnormal coil resistance, calculate the ratio of the duration of the coil current waveform relative to the healthy state, and compare the variation range of the duration under different faults; Compare and analyze the difference between damper failure and spring damage, and compare the extension time of the faulty spring with the contact separation displacement and healthy state for intuitive display; Compare and analyze the differences between coil resistance failure and spring damage, take the maximum value of the relevant current as the decisive feature, and analyze its changes under different faults by drawing a ratio change diagram; According to the difference analysis results between coil voltage abnormality and coil resistance abnormality, damper failure and spring damage, coil resistance failure and spring damage, after the first training of the SVM fault classification diagnosis model, the fault relationship is trained for the second time.

7. The high voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: Also includes: Accelerate the processing of SVM fault classification diagnosis model; The accelerating process of the SVM fault classification diagnosis model includes: To solve the problem of time-consuming training of SVM fault classification diagnosis model, the Laplace operator is used in combination with a parallel acceleration computing module consisting of one-dimensional feature fast transformation to improve the network.

8. The high voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: Also includes: Verify and evaluate the improved SVM fault classification diagnosis model; The verification and evaluation of the improved SVM fault classification diagnosis model includes: The training set data is input into the improved SVM fault classification diagnosis model for training. After the training is completed, a test data set containing a specific number of samples of each category is used for verification testing; The fault diagnosis accuracy is evaluated by the classification results after diagnosis.

9. A high voltage circuit breaker fault diagnosis system, characterized in that: A method for diagnosing a high-voltage circuit breaker fault according to any one of claims 1 to 8, comprising: Waveform analysis module, used to analyze the changing waveforms of contact displacement and coil current when the high-voltage circuit breaker is in operation; A dataset construction module is used to construct a high-voltage circuit breaker fault dataset covering four fault scenarios: coil voltage, coil resistance, damper, and spring structure; Model building module, used to build a preliminary SVM fault classification diagnosis model; A model training module is used to integrate the data relationship between faults and the parallel acceleration module into the preliminary SVM fault classification diagnosis model, and train it to obtain the trained SVM fault classification diagnosis model; The fault diagnosis module is used to diagnose the high-voltage circuit breaker based on the trained SVM fault classification diagnosis model to obtain the fault diagnosis result.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the high-voltage circuit breaker fault diagnosis method according to any one of claims 1 to 8 is implemented.

Images