Support vector machine breaker state discrimination method and system based on improved genetic algorithm optimization, and storage medium

Abstract

The invention discloses a support vector machine circuit breaker state discrimination method and system based on improved genetic algorithm optimization and a storage medium, and the method comprises the steps: S1, obtaining a coil current signal in a circuit breaker, and extracting related feature data; s2, collecting a vibration signal time sequence generated by mechanical interaction in the circuit breaker; s3, designing selection, crossover and variation links of an improved genetic algorithm through a tournament selection mechanism, hybrid crossover and Gaussian variation, and taking a support vector machine discrimination result as a fitness function; s4, selecting a radial basis kernel function, and searching an optimal parameter of a penalty coefficient and kernel function parameter combination based on an improved genetic algorithm to construct an optimized support vector machine model; s5, training the optimized support vector machine model by using the training set data set; and S6, obtaining a current signal and a vibration signal of a circuit breaker coil, extracting features, and inputting the features into the trained and optimized support vector machine model.

Description

Technical Field

[0001] The present application relates to the technical field of power grid system protection, and in particular to a support vector machine circuit breaker state determination method, system and storage medium based on improved genetic algorithm optimization. Background Art

[0002] Circuit breakers are important mechanical switchgear in power grid systems, and their reliability is directly related to the safe and stable operation of smart grids. Circuit breaker failures may cause chaos in system operation and large-scale power outages, and may even cause fires and other personnel safety accidents. Therefore, the technology for distinguishing the state of circuit breakers is of great significance to improving the reliability of circuit breakers and the stability of power grid systems.

[0003] At present, the state identification of circuit breakers mostly relies on low-voltage tests and opening and closing coil current detection. By analyzing the current effective value, peak value, harmonic content and other rich data, various electrical anomalies can be identified. However, relying solely on this signal is prone to misjudgment and other problems, and it is difficult to determine the specific location of the fault. The mechanical waves generated by the circuit breaker during the opening and closing process can directly reflect the mechanical state and movement of the circuit breaker, detect minor mechanical anomalies, and provide early warning. The technologies for circuit breaker state identification methods are mainly divided into four categories: expert system, analytical model, pattern recognition and artificial intelligence. Among them, the expert system knowledge base maintenance investment is large, and it is easy to cause inaccurate state identification results due to the limitations of the knowledge base; the analytical model process is complex and has poor environmental adaptability; in the absence of power grid system data, pattern recognition is prone to local optimality due to data characteristics; deep learning in artificial intelligence has high data requirements and consumes huge computing resources, but the improvement brought to state identification is relatively less obvious.

[0004] Support vector machines are suitable for circuit breaker state discrimination because they are suitable for processing small sample data in power grid systems, have simple models, small computational complexity but high accuracy, and have better adaptability and interpretability. In addition, a few studies have combined genetic algorithms to optimize the parameters of support vector machines, but have ignored the huge amount of computation that genetic algorithms bring to the state discrimination model, which will cause the circuit breaker state discrimination model based on support vector machines to lose its advantage of fast calculation speed. Therefore, it is necessary to design a genetic algorithm suitable for this state discrimination model. Summary of the invention

[0005] In view of the defects existing in the prior art, the present application provides a support vector machine circuit breaker state judgment method, system and storage medium based on improved genetic algorithm optimization to improve the existing circuit breaker detection efficiency and optimize the robustness of circuit breaker state judgment in the power grid system.

[0006] To achieve the above objectives, the present invention provides a method for distinguishing circuit breaker status based on a support vector machine optimized by an improved genetic algorithm, comprising the following steps:

[0007] S1. Obtain the coil current signal in the circuit breaker and extract relevant characteristic data, including current peak value, current change rate, current gradient difference and time difference;

[0008] S2. Collect the vibration signal time series generated by the mechanical interaction inside the circuit breaker, reconstruct the phase space of the sequence, extract relevant features from the phase space, integrate them with the current features according to time, and divide the integrated feature data set into a training set and a test set;

[0009] S3, designing the selection, crossover and mutation links of the improved genetic algorithm through the tournament selection mechanism, hybrid crossover and Gaussian mutation, and using the support vector machine discrimination result as the fitness function;

[0010] S4, selecting a radial basis kernel function, and finding the optimal parameters of the penalty coefficient and the kernel function parameter combination based on an improved genetic algorithm to construct an optimized support vector machine model;

[0011] S5. Use the training set data set to train the optimized support vector machine model, and optimize the model parameters through cross-validation to obtain the best state discrimination performance;

[0012] S6. Obtain the current circuit breaker coil current signal and vibration signal, extract features and input them into the trained and optimized support vector machine model, use the model to classify the data and determine the operating status of the circuit breaker.

[0013] Furthermore, the characteristic data of the optimized support vector machine state discrimination model comes from the current signal and the vibration signal.

[0014] Further, S1, the step of obtaining the coil current signal in the circuit breaker and extracting relevant characteristic data specifically includes:

[0015] Measure the coil current waveform signal of the control circuit during the operation of the circuit breaker tripping mechanism;

[0016] Based on the coil current waveform signal, relevant features are extracted, including: the peak value of the current when the plunger moves, and the corresponding time point t1 and current I1 are extracted as features; the time point t2 and current I2 corresponding to the plunger hitting the latch are extracted as features to reflect the fault of the circuit and the coil; the rate of change of current is extracted as a feature; the current gradient difference before and after the plunger hits the latch is extracted as a feature to characterize the speed change of the plunger; the time point t3 and I3 corresponding to the maximum displacement of the coil plunger are extracted as features, as well as the time difference between t1 and t2, and the time difference from t2 to t3 as features.

[0017] Among them, the current gradient satisfies the relationship:

[0018]

[0019] Where C(t) is the coil current at time t.

[0020] Furthermore, S2 collects the time series of vibration signals generated by the mechanical interaction inside the circuit breaker, reconstructs the sequence in phase space, extracts relevant features from the phase space, integrates them with the current features according to time, and divides the integrated feature data set into a training set and a test set. The vibration signal is divided into three parts: the friction between the plunger and the coil when the plunger starts to move, the shock wave generated when the plunger hits the latch, and the interaction between the operating mechanism and the moving contact when the plunger reaches the maximum displacement.

[0021] Furthermore, the phase space reconstruction of the vibration signal time series is performed, specifically including:

[0022] For a vibration signal time series S that satisfies the relationship S = {s(i), i = 1, 2, ..., n}, the sequence is delayed by introducing a time interval Δt to reconstruct an m-dimensional phase space vector that satisfies the relationship:

[0023]

[0024] Each column represents a phase point in the m-dimensional phase space, and the connecting lines between [n-(m-1)] phase points constitute the trajectory of the system in the phase space, which is called the chaos attraction factor.

[0025] Furthermore, the vibration signal features extracted from the phase space include:

[0026] The minimum limiting sphere radius of the chaotic attractor of the vibration signal can describe the activity range of the vibration signal in the phase space; the maximum distance of the state point in the phase space can characterize the severity of the signal change; the average distance of the state point in the phase space; the ratio of the maximum speed to the average speed; the ratio of the major axis to the minor axis of the minimum outer rectangle of the attractor; the ratio of the attractor coverage area to the minimum enclosing rectangle area.

[0027] 7. Further, S3, the selection, crossover and mutation links of the improved genetic algorithm are designed through the tournament selection mechanism, hybrid crossover and Gaussian mutation, and the support vector machine discrimination result is used as the fitness function. The tournament selection mechanism includes determining the number of individuals participating in the tournament and randomly selecting the individuals of the number from the population for fitness comparison, leaving the individuals with the highest fitness to enter the next generation, and repeating the above operation until the required number of individuals are selected to enter the next generation; the hybrid crossover includes randomly selecting two parent individuals, determining the hybrid factor α, and generating new individuals to satisfy the relationship c = (1-α) * p1 + α * p2, where p1 and p2 are parent individuals, and the order can be exchanged or the hybrid factor can be changed to generate multiple offspring; the Gaussian mutation includes selecting individuals and determining the parameters to be mutated, and generating a new gene X to satisfy the relationship X = x + N (0, σ 2 ), where x is the selected original gene, N(0,σ 2 ) has a mean of 0 and a variance of σ 2 is a Gaussian distributed random variable, σ is the variation intensity parameter, which controls the amplitude of the Gaussian noise.

[0028] Further, S4, a radial basis kernel function is selected, and an optimal parameter combination of a penalty coefficient and a kernel function parameter is found based on an improved genetic algorithm to construct an optimized support vector machine model, wherein the radial basis kernel function satisfies the relationship F(d1, d2) = exp(-γ‖d1-d2‖ 2 ), where d1 and d2 are two data points in the input sample, γ is the kernel function parameter used to control the width of the Gaussian distribution, and ‖d1-d2‖ represents the Euclidean distance between d1 and d2.

[0029] Furthermore, the optimized support vector machine model construction also includes:

[0030] The optimal combination of (C, γ) parameters is determined based on the genetic algorithm, where C is the penalty coefficient. The specific steps are: set C and γ as the composition of each individual and initialize them within the set range; define the fitness function for 5-fold cross-validation, return the average accuracy as the fitness, and maximize the accuracy through the genetic algorithm; perform selection, crossover and mutation genetic operations on the population; through multiple population iterations, each generation updates the population through selection, crossover and mutation, and finally finds the best performing individual, that is, the optimal parameter combination.

[0031] The present invention also provides a circuit breaker state discrimination system based on a support vector machine optimized by an improved genetic algorithm, the system comprising:

[0032] A characteristic data acquisition module, used to acquire a coil current signal and a vibration signal in the circuit breaker, and extract corresponding characteristic data based on the coil current signal and the vibration signal;

[0033] Genetic algorithm design module, used to obtain the optimal combination of (C, γ) parameters, where the selection phase uses the tournament selection mechanism, the crossover phase uses the hybrid crossover strategy, and the mutation phase uses Gaussian mutation;

[0034] A model building module is used to build an optimized support vector machine state discrimination model based on radial basis kernel function and genetic algorithm;

[0035] The model training module uses the extracted feature data training set to train the optimized support vector machine state discrimination model;

[0036] The state discrimination module obtains the current circuit breaker coil current signal and vibration signal, and uses the circuit breaker state discrimination model of the support vector machine based on the improved genetic algorithm optimization to judge the current circuit breaker state.

[0037] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements a method for distinguishing a circuit breaker state based on a support vector machine optimized by an improved genetic algorithm as described in any one of the above.

[0038] Compared with the prior art, the advantages of the present invention are:

[0039] (1) The present invention improves the robustness of circuit breaker state discrimination by extracting coil current signal features and vibration signal features after phase space reconstruction as a source of state discrimination feature data.

[0040] (2) Since the support vector machine algorithm model is simple, the computational complexity is small and the accuracy is high, it is more suitable for processing small sample data in the power grid system. It not only saves computing resources, but also has better adaptability and interpretability than the deep learning model.

[0041] (3) An improved genetic algorithm is designed through a tournament selection mechanism, hybrid crossover, and Gaussian mutation. The optimal combination of (C, γ) parameters is obtained based on the improved genetic algorithm, which can quickly find the optimal solution in irregular data, improve the accuracy of circuit breaker state discrimination based on support vector machine, and maintain its advantage of fast calculation speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 creative work.

[0043] Figure 1A schematic flow chart of a method for distinguishing a circuit breaker state based on a support vector machine optimized by an improved genetic algorithm according to an embodiment of the present application;

[0044] Figure 2 A system block diagram of a circuit breaker state identification system based on a support vector machine optimized by an improved genetic algorithm according to an embodiment of the present application;

[0045] Figure 3 A 3D diagram of a feature data training set of a support vector machine circuit breaker state discrimination model based on improved genetic algorithm optimization according to an embodiment of the present application;

[0046] Figure 4 This is a 3D diagram of a feature data test set of a support vector machine circuit breaker state discrimination model based on improved genetic algorithm optimization according to an embodiment of the present application;

[0047] Figure 5 This is a 3D diagram of the discrimination results of a support vector machine circuit breaker state discrimination model based on improved genetic algorithm optimization in an embodiment of the present application.

[0048] Description of reference numerals:

[0049] 100. Feature data acquisition module; 200. Genetic algorithm design module; 300. Model construction module; 400. Model training module; 500. State discrimination module. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0051] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0052] In the description of the present application, the terms "first / second" involved are merely used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first / second" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0053] It should be understood that the orientation or position relationship is based on the orientation or position relationship shown in the drawings. These orientation terms are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present application.

[0054] Circuit breakers are key protection and control devices in power grid systems, and their normal operation is crucial to the reliability of power supply. Circuit breaker failures, especially electrical failures such as short circuits and overloads, may pose a danger to operators and equipment. Circuit breaker status identification technology can detect and isolate faults in a timely manner, avoid personal injury and equipment damage, and ensure the safety of power grid systems and personnel. Therefore, timely and accurate identification of the circuit breaker status can prevent sudden power outages, ensure the stable operation of the power grid system, and reduce power outages to users.

[0055] In the prior art, there are two key factors that affect the effect of circuit breaker state discrimination. One is the source of the feature data required by the state discrimination model. At present, the feature data of most circuit breaker state discrimination technologies are derived from coil current signals. Although fault states such as poor contact, latch lock jam, and insufficient power supply voltage can be separated in the feature space, the samples in the normal state and the coil-plunger jam fault state are mixed together and difficult to identify. There are also some circuit breaker state discrimination technologies whose feature data are derived from the vibration signal generated by internal mechanical interactions. Although the normal state can be separated from most fault states, the mechanical fault states of coil-plunger jam and open latch jam overlap, resulting in electrical fault states of insufficient power supply and poor contact. When these two features are considered at the same time, the normal state and various fault states can be well separated, which can be identified by machine learning algorithms, which helps to improve the robustness of the circuit breaker state discrimination model. Therefore, this paper first obtains the coil current signal in the circuit breaker and extracts the relevant feature data; and collects the time series of the circuit breaker vibration signal, and then reconstructs the vibration signal time series in phase space to extract the feature data of the vibration signal; then integrates the coil current feature and vibration feature data as the training set data and test set data of the state discrimination model.

[0056] Another key factor affecting the judgment of circuit breaker fault status is the status judgment technology. At present, the technologies for circuit breaker status judgment methods are mainly divided into four categories: expert system, analytical model, pattern recognition and artificial intelligence. Among them, the circuit breaker status judgment technology based on expert system uses predefined rules and reasoning mechanisms to quickly process input information and make judgments, and because its decision-making process is based on clear rules and logic, it is easier to explain its judgment results. However, the performance of the expert system is highly dependent on the quality and integrity of the knowledge base, and the maintenance and update of its knowledge base is relatively large, which can easily lead to inaccurate status judgment results due to the limitations of the knowledge base.

[0057] The circuit breaker state identification technology based on analytical models is based on physical and mathematical principles. It can accurately describe the working state and fault characteristics of the circuit breaker, and clearly understand the generation mechanism and influencing factors of the internal fault of the circuit breaker, which helps to identify the root cause of the fault. However, the construction of the analytical model requires an in-depth understanding of its internal structure, material properties, working principle, etc. This process is complex and redundant, and the established model has poor environmental adaptability and lacks flexibility.

[0058] The circuit breaker state identification technology based on pattern recognition is not limited to sample data and can process a large amount of complex data, including current waveforms, vibration signals, temperature changes, etc. Therefore, it can capture various complex fault characteristics and is suitable for different types of fault state identification. It can also continuously learn by updating training data to improve the accuracy and adaptability of state identification. However, this technology may lead to local optimality in some cases due to data feature problems, especially in the case of data shortage in the power grid system, and it is too dependent on data and converges slowly.

[0059] The circuit breaker state discrimination technology based on artificial intelligence is mainly divided into two types: machine learning and deep learning. They can both identify subtle fault states, so the accuracy of state discrimination is high, and knowledge is extracted from data through automatic learning, which reduces the dependence on manual experience and makes maintenance easier. However, the method based on deep learning requires a large amount of data, and the power grid system usually cannot meet its data needs. In addition, its model is complex and requires high computing resources, which will bring high system costs, and the corresponding discrimination effect is not significantly improved. Therefore, this paper selects the support vector machine algorithm in machine learning, which is more suitable for processing small sample data in the power grid system. The model is simple, the amount of calculation is small, but the accuracy is high, and it has better adaptability and interpretability.

[0060] Genetic algorithm is an optimization algorithm based on natural selection and genetics, and is a type of evolutionary computing. Its core idea is to simulate the biological evolution process and optimize the quality of the solution through operations such as selection, crossover, and mutation. The performance of the support vector machine is affected by hyperparameters, and genetic algorithms can be used to optimize its hyperparameters. They can automatically search the hyperparameter space to reduce the time of manual parameter adjustment, and can find the global optimal parameter combination to avoid local optimality.

[0061] Figure 1 Schematic diagram of a flow chart of a circuit breaker state determination method based on a support vector machine optimized by an improved genetic algorithm according to an embodiment of the present application. Figure 1 As shown, the circuit breaker state discrimination method based on the support vector machine optimized by the improved genetic algorithm includes:

[0062] S1. Obtain the coil current signal in the circuit breaker and extract relevant characteristic data, including current peak value, current change rate, current gradient difference and time difference;

[0063] S2. Collect the vibration signal time series generated by the mechanical interaction inside the circuit breaker, reconstruct the phase space of the sequence, extract relevant features from the phase space, integrate them with the current features according to time, and divide the integrated feature data set into a training set and a test set;

[0064] S3, designing the selection, crossover and mutation links of the improved genetic algorithm through the tournament selection mechanism, hybrid crossover and Gaussian mutation, and using the support vector machine discrimination result as the fitness function;

[0065] S4, selecting a radial basis kernel function, and finding the optimal parameters of the penalty coefficient and the kernel function parameter combination based on an improved genetic algorithm to construct an optimized support vector machine model;

[0066] S5, using the training set data set to train the optimized support vector machine model, and optimizing the model parameters through cross-validation to obtain the best state discrimination performance;

[0067] S6. Obtain the current circuit breaker coil current signal and vibration signal, extract features and input them into the trained and optimized support vector machine model, use the model to classify the data and determine the operating status of the circuit breaker.

[0068] Specifically, the tripping mechanism is a key component of the operating mechanism in the circuit breaker, and the failure of the tripping mechanism is an important cause of the failure of the circuit breaker. The operation process of the tripping mechanism involves the change of the coil current in the control circuit and the movement process of the coil plunger. Therefore, the vibration information of the switch coil current and the plunger movement are both signals that directly reflect the state of the tripping mechanism. First, the coil current waveform of the circuit breaker switching process is obtained by measuring the coil current at each stage of the circuit breaker switching process. Relevant features can be extracted from the current waveform, including: the peak value of the current when the plunger moves, extracting the corresponding time point t1 and current I1 as features; extracting the corresponding time point t2 and current I2 when the plunger hits the latch as features to reflect the failure of the circuit and the coil; extracting the rate of change of current as a feature; extracting the current gradient difference before and after the plunger hits the latch as a feature to characterize the speed change of the plunger; extracting the time point t3 and I3 corresponding to the maximum displacement of the coil plunger as features, as well as the time difference between t1 and t2, and the time difference from t2 to t3 as features.

[0069] In some embodiments, the current gradient characteristic satisfies the relationship:

[0070]

[0071] Where C(t) is the coil current at time t.

[0072] At the same time, the vibration signal of the operating mechanism is measured. The vibration signal is divided into three parts: the friction between the plunger and the coil when the plunger starts to move, the shock wave generated when the plunger hits the latch, and the interaction between the operating mechanism and the moving contact when the plunger reaches the maximum displacement. Since the circuit breaker is a complex nonlinear system, the vibration signal generated during the tripping process is not only a simple output of the plunger hitting the latch, but also contains information about other components and the evolution of the circuit breaker dynamic system. The vibration signal is represented as a one-dimensional chaotic time series in the time domain.

[0073] In some embodiments, the phase space reconstruction of the one-dimensional mixed vibration signal time series specifically includes:

[0074] For a vibration signal time series S that satisfies the relationship S = {s(i), i = 1, 2, ..., n}, the sequence is delayed by introducing a time interval Δt to reconstruct an m-dimensional phase space vector that satisfies the relationship:

[0075]

[0076] Each column represents a phase point in the m-dimensional phase space, and the connecting lines between [n-(m-1)] phase points constitute the trajectory of the system in the phase space, which is called the chaos attraction factor.

[0077] In some embodiments, extracting the vibration signal feature from the phase space includes:

[0078] The minimum limiting sphere radius of the chaotic attractor of the vibration signal can describe the activity range of the vibration signal in the phase space; the maximum distance of the state point in the phase space can characterize the severity of the signal change; the average distance of the state point in the phase space; the ratio of the maximum speed to the average speed; the ratio of the major axis to the minor axis of the minimum outer rectangle of the attractor; the ratio of the attractor coverage area to the minimum enclosing rectangle area.

[0079] After extracting the coil current signal feature data and extracting the feature data from the vibration signal reconstructed in phase space, the coil current signal feature data and the vibration signal feature data are correspondingly integrated according to the time of the circuit breaker switching process to obtain the circuit breaker current-vibration signal feature data set, which is divided into a training set and a test set to prepare for the subsequent circuit breaker state discrimination model.

[0080] After processing the coil current signal and vibration signal to obtain the relevant feature data set, the circuit breaker state discrimination model is constructed based on the support vector machine algorithm. Specifically, due to the complex working principle of the circuit breaker, the state discrimination of the circuit breaker is a nonlinear task. Therefore, the radial basis kernel function is selected, and the optimal parameters of the penalty coefficient and kernel function parameter combination are found based on the genetic algorithm to construct the optimized support vector machine model. The radial basis kernel function satisfies the relationship F(d1, d2) = exp(-γ‖d1-d2‖ 2 ), where d1 and d2 are two data points in the input sample, γ is the kernel function parameter used to control the width of the Gaussian distribution, and ‖d1-d2‖ represents the Euclidean distance between d1 and d2.

[0081] In some embodiments, the support vector machine circuit breaker state determination method based on improved genetic algorithm optimization further includes:

[0082] The selection, crossover and mutation links of the improved genetic algorithm are designed by the tournament selection mechanism, hybrid crossover and Gaussian mutation, and the support vector machine discrimination result is used as the fitness function. The tournament selection mechanism includes determining the number of individuals participating in the tournament and randomly selecting the individuals of the number from the population for fitness comparison, leaving the individuals with the highest fitness to enter the next generation, and repeating the above operation until the required number of individuals are selected to enter the next generation. The application limits the number of individuals to 3 in order to balance the calculation amount and the improvement of the state discrimination accuracy; the hybrid crossover includes randomly selecting two parent individuals, determining the hybrid factor α, and generating new individuals to satisfy the relationship c=(1-α)*p1+α*p2, wherein p1 and p2 are parent individuals, and the order can be exchanged or the hybrid factor can be changed to generate multiple offspring, wherein the hybrid factor ranges from 0 to 1; the Gaussian mutation includes selecting individuals and determining the parameters to be mutated, and generating new genes X to satisfy the relationship X=x+N(0,σ 2), where x is the selected original gene, N(0,σ 2 ) has a mean of 0 and a variance of σ 2 is a Gaussian distributed random variable, σ is the variation intensity parameter, which controls the amplitude of the Gaussian noise.

[0083] In some embodiments, the optimized support vector machine model construction further includes:

[0084] The optimal combination of (C, γ) parameters is determined based on the genetic algorithm, where C is the penalty coefficient. The specific steps are: set C and γ as the composition of each individual and initialize them within the set range; define the fitness function for 5-fold cross-validation, return the average accuracy as the fitness, and maximize the accuracy through the genetic algorithm; perform selection, crossover and mutation genetic operations on the population; through multiple population iterations, each generation updates the population through selection, crossover and mutation, and finally finds the best performing individual, that is, the optimal parameter combination.

[0085] After building the circuit breaker state discrimination model based on the optimized support vector machine, read the saved feature data training set for data preprocessing as input, train the model, and save the model and scaler. Then read the saved feature data test set and load the saved model and scaler, preprocess the read data, use the model for state discrimination, and finally obtain the state discrimination result, and visualize the result.

[0086] In some embodiments, the real-time data of the circuit breaker can be used as the feature data of the test set. When the trained model reads the real-time data of the circuit breaker, the real-time status discrimination result of the circuit breaker can be obtained. And the real-time data can be further supplemented to the model training set data to further train and optimize the model.

[0087] like Figure 2As shown, the embodiment of the present application also provides a support vector machine circuit breaker state discrimination system based on improved genetic algorithm optimization. The support vector machine circuit breaker state discrimination system based on improved genetic algorithm optimization includes a feature data acquisition module 100, a genetic algorithm design module 200, a model construction module 300, a model training module 400 and a state discrimination module 500, wherein the feature data acquisition module is used to obtain the coil current signal and vibration signal in the circuit breaker, and extract the corresponding feature data based on the coil current signal and vibration signal; the genetic algorithm design module is used to obtain the optimal combination of (C, γ) parameters, wherein the selection stage uses a tournament selection mechanism, the crossover stage selects a mixed crossover strategy, and the mutation stage uses Gaussian mutation; the model construction module is used to construct an optimized support vector machine state discrimination model based on a radial basis kernel function and a genetic algorithm; the model training module uses the extracted feature data training set to train the optimized support vector machine state discrimination model; the state discrimination module obtains the current circuit breaker coil current signal and vibration signal, and uses the support vector machine circuit breaker state discrimination model based on the improved genetic algorithm optimization to determine the current circuit breaker state.

[0088] In order to better understand the support vector machine circuit breaker state discrimination method based on improved genetic algorithm optimization in the embodiment of the present application, Figure 3 , Figure 4 and Figure 5 Provide a detailed description.

[0089] This application takes the time difference between the current reaching the peak value and the time when the plunger hits the latch, the current gradient difference before and after the plunger hits the latch, and the average distance of the state point in the phase space as examples, measures the coil current at each stage of the circuit breaker switching process to obtain the coil current waveform of the circuit breaker switching process, and extracts relevant features from the current waveform. The vibration signal of the measured operating mechanism obtains a one-dimensional chaotic time series of the vibration signal, reconstructs the one-dimensional chaotic vibration signal time series in phase space, and extracts relevant features.

[0090] The extracted current signal-vibration signal features are integrated to obtain feature sample data under different states, and divided into training set data and test set data, such as Figure 3 , Figure 4As shown, the three dimensions represent the three features in the example, namely, the time difference feature between the current reaching the peak value and the time when the plunger hits the latch, the current gradient difference feature before and after the plunger hits the latch, and the average distance feature of the state point in the phase space. The solid points of different colors in the three-dimensional space represent sample data in different states. In this example, there are four states, namely normal state, fault 1, fault 2 and fault 3. This example is only used to help understand the support vector machine circuit breaker state discrimination method based on improved genetic algorithm optimization. This method can also be extended to more categories of feature data and more categories of fault states. Next, regarding the design of the improved genetic algorithm, the selection scale can be controlled through the tournament selection mechanism to control the algorithm complexity. The application selects a smaller scale to enhance the dominant individuals selected, which can not only save computing resources but also prevent the population from converging to the local optimum too early; hybrid crossover can produce smoother offspring changes by weighted combination of parent individuals, which is helpful to explore continuous space, and can retain multiple features of the parent generation, which is helpful to maintain population diversity and reduce the useful information lost by crossover; Gaussian mutation can effectively enhance the diversity of the population by applying small random perturbations, and can reduce the risk of introducing extreme solutions. In addition, the circuit breaker state discrimination model is constructed based on the optimized support vector machine algorithm. The radial basis kernel function is selected, and C and γ are set as the composition of each individual based on the improved genetic algorithm, and initialized within the set range; and the fitness function is defined for 5-fold cross validation, and the average accuracy is returned as the fitness, and the accuracy is maximized through the improved genetic algorithm; the population undergoes selection, crossover and mutation genetic operations; through multiple population iterations, each generation updates the population through selection, crossover and mutation, and finally finds the best performing individual, that is, the optimal parameter combination, where C is the penalty coefficient and γ is the kernel function parameter. After the circuit breaker state discrimination model based on the optimized support vector machine is constructed, the saved feature data training set is read for data preprocessing as input, the model is trained, and the model and scaler are saved. After the model training is completed, the saved feature data test set is read and the saved model and scaler are loaded, the read data is preprocessed, and the model is used for state discrimination, and finally the state discrimination result is obtained, and the result is visualized, as shown in the figure. Figure 5 As shown. Figure 4 and Figure 5 It can be found that the state discrimination accuracy of the support vector machine circuit breaker state discrimination method based on improved genetic algorithm optimization proposed in this application reaches 95%. Similarly, the real-time data of the circuit breaker can also be used as the test set feature data. When the trained model reads the real-time data of the circuit breaker, the real-time state discrimination result of the circuit breaker can be obtained. And the real-time data can be further supplemented to the model training set data to further train and optimize the model.

[0091] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned support vector machine circuit breaker state determination method based on improved genetic algorithm optimization is implemented.

[0092] It should be noted that the storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a ROM (Read-Only Memory), a RAM (Random Access Memory), a magnetic disk or an optical disk.

[0093] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable storage medium, which may include a computer-readable storage medium (or a non-transitory medium) and a communication medium (or a temporary medium).

[0094] As is known to those of ordinary skill in the art, the term computer-readable storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer-readable storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is known to those of ordinary skill in the art that communication media typically contain computer-readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and may include any information delivery medium.

[0095] Exemplarily, the computer-readable storage medium may be an internal storage unit of the electronic device of the aforementioned embodiment, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device.

[0096] The above is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features applied for herein.

Claims

1. A method for distinguishing circuit breaker status based on support vector machine optimized by improved genetic algorithm, characterized in that: The steps include: S1. Obtain the coil current signal in the circuit breaker and extract relevant characteristic data, including current peak value, current change rate, current gradient difference and time difference; S2. Collect the vibration signal time series generated by the mechanical interaction inside the circuit breaker, reconstruct the phase space of the sequence, extract relevant features from the phase space, integrate them with the current features according to time, and divide the integrated feature data set into a training set and a test set; S3, designing the selection, crossover and mutation links of the improved genetic algorithm through the tournament selection mechanism, hybrid crossover and Gaussian mutation, and using the support vector machine discrimination result as the fitness function; S4, selecting a radial basis kernel function, and finding the optimal parameters of the penalty coefficient and the kernel function parameter combination based on an improved genetic algorithm to construct an optimized support vector machine model; S5. Use the training set data set to train the improved support vector machine model, and optimize the model parameters through cross-validation to obtain the best state discrimination performance; S6. Obtain the current circuit breaker coil current signal and vibration signal, extract features and input them into the trained and optimized support vector machine model, use the model to classify the data and determine the operating status of the circuit breaker.

2. The circuit breaker state identification method based on support vector machine optimized by improved genetic algorithm according to claim 1 is characterized in that: The characteristic data of the optimized support vector machine state discrimination model comes from the current signal and the vibration signal.

3. The circuit breaker state identification method based on support vector machine optimized by improved genetic algorithm according to claim 1 is characterized in that: S1, the step of obtaining the coil current signal in the circuit breaker and extracting relevant characteristic data specifically includes: Measure the coil current waveform signal of the control circuit during the operation of the circuit breaker tripping mechanism; Based on the coil current waveform signal, relevant features are extracted, including: the peak value of the current when the plunger moves, and the corresponding time point t1 and current I1 are extracted as features; the time point t2 and current I2 corresponding to the plunger hitting the latch are extracted as features to reflect the fault of the circuit and the coil; the rate of change of current is extracted as a feature; the current gradient difference before and after the plunger hits the latch is extracted as a feature to characterize the speed change of the plunger; the time point t3 and I3 corresponding to the maximum displacement of the coil plunger are extracted as features, as well as the time difference between t1 and t2, and the time difference from t2 to t3 as features. Among them, the current gradient satisfies the relationship: Where C(t) is the coil current at time t.

4. The circuit breaker state identification method based on support vector machine optimized by improved genetic algorithm according to claim 1 is characterized in that: S2. Collect the time series of vibration signals generated by the mechanical interaction inside the circuit breaker, reconstruct the phase space of the sequence, extract relevant features from the phase space, and integrate them with the current features according to time, and divide the integrated feature data set into a training set and a test set. The vibration signal is divided into three parts: the friction between the plunger and the coil when the plunger starts to move, the shock wave generated when the plunger hits the latch, and the interaction between the operating mechanism and the moving contact when the plunger reaches the maximum displacement.

5. The circuit breaker state identification method based on support vector machine optimized by improved genetic algorithm according to claim 4 is characterized in that: The phase space reconstruction of the vibration signal time series includes: For a vibration signal time series S that satisfies the relationship S = {s(i), i = 1, 2, ..., n}, the sequence is delayed by introducing a time interval Δt to reconstruct an m-dimensional phase space vector that satisfies the relationship: Each column represents a phase point in the m-dimensional phase space, and the connecting lines between [n-(m-1)] phase points constitute the trajectory of the system in the phase space, which is called the chaos attraction factor. The vibration signal features extracted from the phase space include: The minimum limiting sphere radius of the chaotic attractor of the vibration signal can describe the activity range of the vibration signal in the phase space; the maximum distance of the state point in the phase space can characterize the severity of the signal change; the average distance of the state point in the phase space; the ratio of the maximum speed to the average speed; the ratio of the major axis to the minor axis of the minimum outer rectangle of the attractor; the ratio of the attractor coverage area to the minimum enclosing rectangle area.

6. The circuit breaker state identification method based on support vector machine optimized by improved genetic algorithm according to claim 1, characterized in that: S3. The selection, crossover and mutation links of the improved genetic algorithm are designed through the tournament selection mechanism, hybrid crossover and Gaussian mutation, and the support vector machine discrimination result is used as the fitness function. The tournament selection mechanism includes determining the number of individuals participating in the tournament and randomly selecting the individuals of the number from the population for fitness comparison, leaving the individuals with the highest fitness to enter the next generation, and repeating the above operation until the required number of individuals are selected to enter the next generation (selecting 3 is to balance the amount of calculation, algorithm complexity and accuracy improvement effect, and the effect can be described in detail later); the hybrid crossover includes randomly selecting two parent individuals, determining the hybrid factor α, and generating new individuals to satisfy the relationship c=(1-α)*p1+α*p2, where p1 and p2 are parent individuals, and the order can be exchanged or the hybrid factor can be changed to generate multiple offspring; (the details and advantages such as the range of the hybrid factor will be described in detail later); the Gaussian mutation includes selecting individuals and determining the parameters that need to be mutated, and the new gene X is generated to satisfy the relationship X=x+N(0,σ 2 ), where x is the selected original gene, N(0,σ 2 ) has a mean of 0 and a variance of σ 2 is a Gaussian distributed random variable, σ is the variation intensity parameter, which controls the amplitude of the Gaussian noise.

7. The circuit breaker state identification method based on support vector machine optimized by improved genetic algorithm according to claim 1 is characterized in that: S4, select the radial basis kernel function, and find the optimal parameters of the penalty coefficient and the kernel function parameter combination based on the improved genetic algorithm to construct an optimized support vector machine model, wherein the radial basis kernel function satisfies the relationship F(d1, d2) = exp(-γ‖d1-d2‖ 2 ), where d1 and d2 are two data points in the input sample, γ is the kernel function parameter used to control the width of the Gaussian distribution, and ‖d1-d2‖ represents the Euclidean distance between d1 and d2.

8. The circuit breaker state identification method based on support vector machine optimized by improved genetic algorithm according to claim 7 is characterized in that: The optimized support vector machine model construction also includes: Based on the improved genetic algorithm, the optimal combination of (C, γ) parameters is determined, where C is the penalty coefficient. The specific steps are: set C and γ as the composition of each individual and initialize them within the set range; define the fitness function for 5-fold cross-validation, return the average accuracy as the fitness, and maximize the accuracy through the genetic algorithm; perform selection, crossover and mutation genetic operations on the population; through multiple population iterations, each generation updates the population through selection, crossover and mutation, and finally finds the best performing individual, that is, the optimal parameter combination.

9. A circuit breaker state identification system based on support vector machine optimized by improved genetic algorithm, characterized in that: The system comprises: A characteristic data acquisition module, used to acquire a coil current signal and a vibration signal in the circuit breaker, and extract corresponding characteristic data based on the coil current signal and the vibration signal; Genetic algorithm design module, used to obtain the optimal combination of (C, γ) parameters, where the selection phase uses the tournament selection mechanism, the crossover phase uses the hybrid crossover strategy, and the mutation phase uses Gaussian mutation; A model building module is used to build an optimized support vector machine state discrimination model based on radial basis kernel function and improved genetic algorithm; The model training module uses the extracted feature data training set to train the optimized support vector machine state discrimination model; The state discrimination module obtains the current circuit breaker coil current signal and vibration signal, and uses the circuit breaker state discrimination model of the support vector machine based on the improved genetic algorithm optimization to judge the current circuit breaker state.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for determining the state of a circuit breaker based on a support vector machine optimized by an improved genetic algorithm as claimed in any one of claims 1 to 8 is implemented.

Images