Circuit breaker fault detection optimization method and system based on convolutional neural network

Abstract

The invention discloses a circuit breaker fault detection optimization method and system based on a convolutional neural network. The method comprises the steps of obtaining an original vibration signal generated when a circuit breaker is in an operation state; preprocessing the original vibration signal, and performing feature extraction on the preprocessed original vibration signal based on a convolutional neural network to obtain an extracted target feature data set; constructing an MTBO-CNN model, taking the CNN as a basic model, and introducing an MTBO algorithm to automatically search for optimal parameter configuration; using the training set to train an MTBO-CNN model, and obtaining a trained MTBO-CNN model; and inputting the extracted target feature data into the trained MTBO-CNN model, and classifying the circuit breaker vibration signals by using the optimized network to obtain a circuit breaker fault identification result, so that the problem of instability of the traditional CNN model can be solved, the optimal parameters of the CNN can be stably searched based on the MTBO algorithm, and the performance stability of the system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical equipment maintenance, and in particular to a circuit breaker fault detection optimization method and system based on convolutional neural network. Background Art

[0002] High-voltage circuit breakers are essential components in the main electrical wiring of power systems, and they play a key role in maintaining the stability, reliability, and economy of power systems. They have two main functions: one is to separate and connect normal currents to cope with changes in grid load and ensure the normal operation of the circuit; the other is to cut off fault currents to isolate fault areas and protect the safe operation of power systems. When high-voltage circuit breakers lose reliability or mechanical parts are damaged, two major problems will arise. On the one hand, it may cause instability in the power system or power equipment. On the other hand, it may cause the relay protection to fail to trigger accurately and lose the ability to respond quickly. This will trigger the operation of the upper level protection, expand the scope of the fault, lose selectivity, and in severe cases may cause the entire system to be decoupled.

[0003] At present, the fault diagnosis of circuit breakers mainly includes wavelet transform, empirical mode decomposition, Hilbert transform and envelope method for diagnostic signals, followed by fault classification using support vector machine, random forest, neural network and other algorithms. In fault diagnosis, the presence of noise and outliers may have a negative impact on the accuracy of the model. High-dimensional sparse data will make it difficult to split the nodes of the decision tree, and more samples may be required for effective training and modeling. Compared with support vector machines, convolutional neural networks (CNNs) usually have better modeling capabilities when dealing with nonlinear problems. CNNs can learn complex nonlinear relationships by using nonlinear activation functions and multi-layer network structures, thereby better adapting to the modeling needs of nonlinear problems. However, since the parameters of the neural network need to be determined manually, the selection of hyperparameters is crucial. Once the selection is inaccurate, it is easy to fall into a local optimal solution. The performance of CNN is highly dependent on its parameter settings, and effective parameter optimization is required. Traditional CNN methods usually use randomly initialized parameters, which may lead to poor model performance and require long-term training and manual parameter adjustment to obtain satisfactory performance, which increases system complexity and computational overhead. Current technology relies on domain experts to select and adjust CNN parameter settings, which limits its wide application because not every operator has expertise in the relevant field. Summary of the invention

[0004] The purpose of the present invention is to provide a circuit breaker fault detection optimization method, device and equipment based on convolutional neural network, aiming to solve the instability problem of traditional CNN model, and to stably search for the optimal parameters of CNN based on MTBO algorithm to improve the performance stability of the system.

[0005] The present invention provides a circuit breaker fault detection optimization method based on convolutional neural network, comprising:

[0006] Acquire the original vibration signal generated when the circuit breaker is in operation;

[0007] Preprocessing the original vibration signal, and extracting features of the preprocessed original vibration signal based on a convolutional neural network to obtain an extracted target feature data set;

[0008] Construct the MTBO-CNN model, using the CNN convolutional neural network as the basic model, and introduce the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model;

[0009] Dividing the target feature data set into a training set and a test set according to a preset ratio, and using the training set to train the MTBO-CNN model, and using the test set to test the MTBO-CNN model, to obtain a trained MTBO-CNN model;

[0010] The extracted target feature data is input into the trained MTBO-CNN model, and the optimized network is used to classify the circuit breaker vibration signal to obtain the circuit breaker fault identification result.

[0011] Preferably, the obtaining of the original vibration signal generated when the circuit breaker is in operation further comprises:

[0012] Construct a vibration signal acquisition system with a circuit breaker, set a vibration sensor on the circuit breaker, and start a working operation mode to simulate different mechanical fault types of the circuit breaker;

[0013] Based on the vibration signal acquisition system, the signal acquisition device is activated to capture the analog vibration signal of the circuit breaker in the mechanical state, and the captured analog vibration signal is converted into a digital signal, which is collected and recorded according to a preset sampling rate, and the collected analog vibration signal is transmitted to the fault diagnosis system to perform circuit breaker fault detection operation;

[0014] Among them, the vibration signal acquisition system includes a circuit breaker, a signal acquisition device and a fault diagnosis system connected in sequence, and the signal acquisition device is powered by an external power supply to ensure the normal operation of the acquisition work; the simulated vibration signal includes loose screws on the base of the circuit breaker, stuck iron core, insufficient lubrication and related fault signals of different degrees simulated by adjusting the corresponding mechanical parts on the circuit breaker.

[0015] Preferably, in constructing the MTBO-CNN model, a CNN convolutional neural network is used to perform signal processing and feature extraction in sequence according to the operation sequence of the neural network, which can efficiently realize multi-level information fusion. The CNN convolutional neural network includes an input layer, a one-dimensional convolutional layer, a two-dimensional convolutional layer, a correction layer, a maximum pooling layer and a fully connected layer connected in sequence. The original vibration signal is received from the test source output as the input of the input layer, and is passed to the next network layer layer by layer to output the fault identification result.

[0016] Preferably, the introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model includes:

[0017] The parameter combination of the CNN convolutional neural network is defined as θ = {α, k, n k}, to obtain the best parameter combination θ*, so that the objective function F(θ) on the validation set is maximized as the goal, and the objective function is set as a weighted combination of classification accuracy and training time

[0018] F(θ)=Accuracy(θ)-λTraining Time(θ);

[0019] Among them, α represents the learning rate, k represents the convolution kernel size, and n k represents the number of convolution kernels; Accuracy(θ) is the classification accuracy under the parameter combination θ, Training Time(θ) is the training time under the parameter combination θ, and λ is the time penalty coefficient;

[0020] Based on the mountaineering team optimization algorithm, each parameter in the hyperparameter is initialized, and a mountaineering team is defined. Each mountaineering team includes multiple climbers, and each climber corresponds to an initial parameter combination θ i ={α i , k i , n k，i}, where i is the subgroup index;

[0021] Take each initial parameter combination θi as the hyperparameter of the CNN model, train the CNN model, calculate the corresponding accuracy Accuracy(θ) on the validation set, record the training time Training Time(θ), calculate the objective function F(θ), and evaluate and optimize the initial parameter combination θi according to the calculation results.

[0022] Preferably, the introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model further includes:

[0023] Based on the search strategy corresponding to MTBO, each sub-population is searched and updated according to the current fitness;

[0024] If a local search is performed, the gradient is calculated for each initial parameter combination:

[0025]

[0026] in Represents the gradient of the current initial parameter combination; Represents the partial derivative of the objective function with respect to the learning rate α; is the partial derivative of the objective function with respect to the convolution kernel size k; The objective function is the number of convolution kernels n k The partial derivative of

[0027] The parameter combination after local search is:

[0028]

[0029] Among them, η is the step size parameter; θi is the current initial parameter combination, θ i-local is the parameter combination after local search;

[0030] If cross-group learning is performed, the parameters are adjusted through information sharing. Assuming that the best parameter combination for fitness in the tth generation is θ*, then:

[0031] θ i-cross =γ·θ*+(1-γ)·θ i-local ;

[0032] Among them, γ is the information reference factor; θ* is the parameter combination with the best fitness in the tth generation; θ i-cross is the parameter combination after cross-group learning.

[0033] Preferably, the introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model further includes:

[0034] Comprehensively combine the parameter combinations after cross-group learning and select the optimal parameter combination θ best :

[0035] θ best = argmaxF(θ i-cross )

[0036] Among them, argmax represents the independent variable that makes the function reach the maximum value, F(θ i-cross ) is the objective function to be maximized;

[0037] Introducing random mutations to maintain the diversity of the search, the updated new parameter combination θ(i,new) is expressed as follows:

[0038] θ i，new =θ i-cross +∈·ramdn();

[0039] Where, ∈ is the variation intensity; randn() is the standard normal random noise;

[0040] The updated new parameter combination θ(i,new) replaces the original θi as the starting point for the next round of iteration.

[0041] Preferably, the introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model further includes:

[0042] Determine whether the above optimization process meets the following termination conditions:

[0043] Determine whether the maximum number of iterations Tmax has been reached; and / or,

[0044] According to the comparison between the best fitness of the current t generation and the best fitness of the t-1 generation, it is judged whether the corresponding fitness value of the current optimal combination has been improved, that is:

[0045]

[0046] In the formula, δ is the convergence threshold, is the optimization result under the t-th generation optimal parameter combination θbest, i.e., the fitness value;

[0047] If the above termination conditions are met, the optimization process is terminated and the final optimal parameter combination configuration θ is determined. best And train the final model;

[0048] If the above termination conditions are not met, the process returns to the step of evaluating and optimizing the initial parameter combination θi and continues the iteration.

[0049] The present invention also provides a circuit breaker fault detection optimization system based on convolutional neural network, comprising:

[0050] A data acquisition module, used for acquiring an original vibration signal generated when the circuit breaker is in operation;

[0051] A preprocessing module, used to preprocess the original vibration signal, and extract features from the preprocessed original vibration signal based on a convolutional neural network to obtain an extracted target feature data set;

[0052] MTBO-CNN model, used to build the MTBO-CNN model, using the CNN convolutional neural network as the basic model, and introducing the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model;

[0053] A model training module, used for dividing the target feature data set into a training set and a test set according to a preset ratio, and using the training set to train the MTBO-CNN model, and using the test set to test the MTBO-CNN model, to obtain a trained MTBO-CNN model;

[0054] The fault detection module is used to input the extracted target feature data into the trained MTBO-CNN model, and use the optimized network to classify the circuit breaker vibration signal to obtain the circuit breaker fault identification result.

[0055] The present invention also provides an electronic device, comprising:

[0056] A memory, the memory being used to store a processing program;

[0057] A processor, wherein when executing the processing program, the processor implements the circuit breaker fault detection optimization method based on convolutional neural network as described in the embodiment of the present invention.

[0058] The present invention also provides a readable storage medium, on which a processing program is stored. When the processing program is executed by a processor, the circuit breaker fault detection optimization method based on a convolutional neural network as described in an embodiment of the present invention is implemented.

[0059] With respect to the prior art, the present invention has the following beneficial effects:

[0060] The present invention proposes a circuit breaker fault detection optimization method based on convolutional neural network. The MTBO algorithm is used to automatically search for the optimal parameter configuration of CNN, which significantly reduces the need for manual parameter adjustment and improves the efficiency of model adjustment. The application of the MTBO algorithm ensures that the optimal solution is quickly and effectively found in a wide parameter space, avoiding the inefficiency and blindness that may occur in the traditional trial and error method. The MTBO-CNN model is used to learn a large amount of vibration signal data, automatically extract key features, and perform effective pattern recognition, so that various fault types of the circuit breaker can be accurately identified. Compared with traditional fault detection methods, this method has higher accuracy and lower false alarm rate. Since the MTBO algorithm considers the performance of the model on the verification set during the parameter optimization process, and maintains the breadth and depth of the search through diversified search strategies, the obtained MTBO-CNN model not only performs well on the training set, but also has good generalization ability, and can effectively deal with new fault types. The automated parameter optimization and efficient fault detection process greatly reduce the dependence on professionals and reduce labor costs. At the same time, fast and accurate fault diagnosis also reduces equipment downtime and improves the overall efficiency of the system. It not only improves the accuracy of circuit breaker fault detection, but also reduces manual intervention through automated parameter adjustment, thereby improving the efficiency and reliability of fault detection.

[0061] This paper takes the vibration signal of the circuit breaker as the research object and proposes a mountaineering team optimization algorithm to optimize the convolutional neural network (MTBO-CNN). The algorithm simulates the behavior of the mountaineering team and aims to find the global optimal solution. Specifically, MTBO optimizes the performance of the model by adjusting the hyperparameters of CNN, such as the convolution kernel size, the number of convolution layers, the activation function, etc. Through experimental verification, MTBO-CNN has achieved significant improvements in the fault diagnosis of circuit breaker vibration signals. Compared with the traditional method of manually determining parameters, MTBO-CNN can automatically search for the optimal hyperparameter combination, thereby improving the performance and generalization ability of the model. In addition, MTBO-CNN also has good robustness and can achieve good results on different data sets and problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1-2 Schematic diagram of the steps of a circuit breaker fault detection optimization method based on a convolutional neural network according to an embodiment of the present invention;

[0063] Figure 3 This is a principle block diagram of a vibration signal acquisition system in one embodiment of the present invention;

[0064] Figure 4 This is an example diagram of a signal acquisition device in one embodiment of the present invention;

[0065] Figure 5This is an example diagram of the steps of the optimization algorithm of MTBO-CNN in one embodiment of the present invention;

[0066] Figure 6 An example diagram of a neural network in one embodiment of the present invention;

[0067] Figure 7 This is an example diagram of a convolutional layer in one embodiment of the present invention;

[0068] Figure 8 This is an example diagram of a convolution operation in a one-dimensional convolution layer in one embodiment of the present invention;

[0069] Fig. 9 This is an example diagram of a convolution operation in a two-dimensional convolutional layer in one embodiment of the present invention;

[0070] Fig.10 This is a signal diagram of a normal state of an original vibration signal in an embodiment of the present invention;

[0071] Fig.11 This is a signal diagram of the original vibration signal stuck state in one embodiment of the present invention;

[0072] Fig.12 This is a signal diagram of the screw loosening state of the original vibration signal in one embodiment of the present invention;

[0073] Fig.13 It is a signal diagram of the lubrication insufficient state of the original vibration signal in one embodiment of the present invention;

[0074] Fig.14 It is a graph showing iterative error variation in one embodiment of the present invention;

[0075] Fig.15 This is an example diagram of a confusion matrix used in an embodiment of the present invention to quantitatively analyze diagnosis results. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0078] It should be noted that the concepts such as "once" and "second" mentioned in the present application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0079] It should be noted that the modifications of "one" and "plurality" mentioned in the disclosure of this application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0080] Embodiment 1

[0081] like Figure 1 As shown, the present invention also provides a circuit breaker fault detection optimization method based on convolutional neural network, comprising:

[0082] S1: Acquire the original vibration signal generated when the circuit breaker is in operation;

[0083] S2: preprocessing the original vibration signal, extracting features from the preprocessed original vibration signal based on a convolutional neural network to obtain an extracted target feature data set;

[0084] S3: Construct the MTBO-CNN model, using the CNN convolutional neural network as the basic model, and introduce the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model;

[0085] S4: dividing the target feature data set into a training set and a test set according to a preset ratio, and using the training set to train the MTBO-CNN model, and using the test set to test the MTBO-CNN model, to obtain a trained MTBO-CNN model;

[0086] S5: The extracted target feature data is input into the trained MTBO-CNN model, and the optimized network is used to classify the circuit breaker vibration signal to obtain the circuit breaker fault identification result.

[0087] This embodiment first collects the vibration signal of the circuit breaker when it is working. To ensure the reliability and quality of the data, the collected signals undergo necessary preprocessing steps, such as denoising, normalization, and feature extraction. The processed vibration signal data set is divided into a training set and a test set in a ratio of 8:2. The training set is used to train the MTBO-CNN network model to help the model learn the complex relationship between vibration signals and specific fault types; the test set is used to evaluate the generalization ability and recognition accuracy of the model. Subsequently, these preprocessed data are input into a deep learning architecture that combines the MTBO (Mountain Team-based Optimization) optimization algorithm and a convolutional neural network (CNN). The MTBO algorithm is used to automatically adjust the hyperparameters of CNN to obtain better model performance. Through multiple iterations and cross-subgroup learning, the MTBO-CNN model can effectively improve the accuracy and efficiency of fault identification. The overall process is as follows: Figure 2 shown.

[0088] In this embodiment, the step S1 of acquiring the original vibration signal generated when the circuit breaker is in operation further includes:

[0089] Construct a vibration signal acquisition system with a circuit breaker, set a vibration sensor on the circuit breaker, and start a working operation mode to simulate different mechanical fault types of the circuit breaker;

[0090] Based on the vibration signal acquisition system, the signal acquisition device is activated to capture the analog vibration signal of the circuit breaker in the mechanical state, and the captured analog vibration signal is converted into a digital signal, which is collected and recorded according to a preset sampling rate, and the collected analog vibration signal is transmitted to the fault diagnosis system to perform circuit breaker fault detection operation;

[0091] Among them, the vibration signal acquisition system includes a circuit breaker, a signal acquisition device and a fault diagnosis system connected in sequence, and the signal acquisition device is powered by an external power supply to ensure the normal operation of the acquisition work; the simulated vibration signal includes loose screws on the base of the circuit breaker, stuck iron core, insufficient lubrication and related fault signals of different degrees simulated by adjusting the corresponding mechanical parts on the circuit breaker.

[0092] In order to effectively detect circuit breaker faults, this embodiment designs and builds a high-precision vibration signal acquisition system, combined with advanced algorithm optimization to improve the accuracy and efficiency of fault identification. The specific acquisition and identification process is as follows: Figure 3, Step 1, power supply of the signal acquisition device: First, the signal acquisition device is powered by an external power supply to ensure that the power supply voltage and current are stable to ensure the normal operation of the acquisition device. Step 2, configuration of the vibration sensor: stick the vibration sensor above the arc extinguishing chamber of the circuit breaker. Step 3, vibration signal acquisition: activate the signal acquisition device and start collecting vibration data. The acquisition device converts the analog vibration signal into a digital signal through the built-in analog-to-digital converter (ADC) and records it at a predetermined sampling rate such as 25.6kHz. Step 4, data input to the fault diagnosis system: the collected vibration signal data is transmitted to the fault diagnosis system by wire. Step 5, fault data analysis and identification: the data input to the fault diagnosis system is first filtered, denoised and signal enhanced through the preprocessing module to highlight the fault characteristics, and then the fault diagnosis algorithm proposed in this embodiment is used for state identification. Step 6, identification result display: the analysis results are displayed through the user interface, and the user can directly observe the status information of the circuit breaker.

[0093] The specific construction of the vibration signal acquisition system in this embodiment is as follows:

[0094] A vibration signal acquisition system was built using PCB Piezotronics accelerometer 356A04 piezoelectric acceleration sensor and NI 9234 data acquisition card produced by National Instruments (NI).

[0095] (1) Sensor selection: Since the vibration signal of the circuit breaker has the characteristics of short effective time and large vibration impact, the CT1005LC acceleration sensor is selected. The sensor is fixed above the arc extinguishing of the circuit breaker by magnet adsorption to ensure signal repeatability and easy installation. With its high-performance SMR sensitive element and precise internal amplification circuit, the sensor has high resolution and high sensitivity. (2) Signal acquisition: The sensor reads ADC data through a parallel interface with a sampling frequency of up to 1MHz. The collected data is downsampled to 32kHz by the FPGA module to make it more suitable for subsequent signal processing and analysis, see Figure 4 (3) Collection settings: The coordinate zero point of the sampling signal is set to the time when the circuit breaker is about to operate to ensure that the data collected each time is consistent. Vibration signals are captured under four mechanical conditions, including loose base screws, core jamming, and insufficient lubrication. Different degrees of faults are simulated by adjusting the conditions of related mechanical components.

[0096] The optimization algorithm based on MTBO-CNN in this embodiment optimizes the three key parameters of the convolutional neural network, namely the learning rate, convolution kernel size, and the number of convolution kernels, through the mountaineering team optimization algorithm, with the highest test set accuracy as the objective function, so as to obtain better performance and more accurate results in the network training process. In this optimization process, the mountaineering team algorithm is different from the traditional optimization algorithm. It simulates the process of a mountaineering team climbing a mountain, and continuously explores different points and directions in the parameter space to find better parameter combinations. Specifically, this algorithm first randomly initializes a set of parameter combinations as the starting point and calculates the corresponding test set accuracy. Then, it searches for nearby parameter points through slight changes and adjustments, and evaluates the improvement in test set accuracy at each step. If a better parameter combination is found, the mountaineering team algorithm will update the parameters and further explore the surrounding neighborhood. This process will continue until a local optimal solution is found or the preset termination condition is reached. The algorithm flow chart is as follows Figure 5 shown.

[0097] In the CNN recognition network of this embodiment, the fault diagnosis system adopts a convolutional neural network (CNN), which includes six key layers, namely the input layer, the one-dimensional convolution layer, the two-dimensional convolution layer, the correction layer, the maximum pooling layer and the full connection layer. These layers perform signal processing and feature extraction in sequence according to the operation sequence of the neural network, and can efficiently realize multi-level information fusion, thereby improving the accuracy of fault diagnosis. The schematic diagram of the structure is shown in FIG. Figure 6 As shown. In the construction of the MTBO-CNN model, the CNN convolutional neural network is used to perform signal processing and feature extraction in sequence according to the operation sequence of the neural network, which can efficiently realize multi-level information fusion. The CNN convolutional neural network includes an input layer, a one-dimensional convolutional layer, a two-dimensional convolutional layer, a correction layer, a maximum pooling layer and a full connection layer connected in sequence. The original vibration signal is received from the test source output as the input of the input layer, and is passed to the next network layer layer by layer to output the fault identification result.

[0098] (1) Input stage

[0099] In the input stage, the system first receives the raw data of the vibration signal from the test source output and performs preprocessing. The data is arranged in time sequence, and a data sample is input every 16 cycles. This process ensures the orderliness of the input data and lays the foundation for subsequent convolution processing. Once the data is collected, it is gradually passed to the next network layer for further processing. The convolution layer consists of multiple convolution cores, each of which consists of two multipliers and an adder. Its structure is as follows Figure 7 .

[0100] (2) One-dimensional convolutional layer

[0101] The one-dimensional convolution layer consists of 16 parallel convolution cores. After each input data item is processed by this layer, 16 different output data can be generated at the same time. The convolution operation is as follows Figure 8 This design enhances the density and effectiveness of feature extraction and improves the quality of the preliminary analysis stage.

[0102] (3) 2D convolutional layer

[0103] The two-dimensional convolution layer convolves the 16 data (corrected) output by the one-dimensional convolution layer. Since the 16 data need to be calculated by 16 convolution cores respectively, in order to save resources, the 16 sets of convolution parameters are input into the 16 convolution cores in a cycle, and a two-dimensional convolution calculation is completed in 16 clock cycles, as shown in the figure. Fig. 9 shown.

[0104] (4) Correction layer

[0105] A correction operation is performed on the output of the convolutional layer, where negative values ​​are set to 0 to ensure that all output data is non-negative, enhancing the interpretability of the model and the stability of subsequent processing.

[0106] (5) Maximum pooling layer

[0107] A pooling operation is applied to the output of the 2D convolutional layer to extract the maximum value from each convolution output channel. Finally, the maximum value of 16 channels is summarized as the output of the pooling layer.

[0108] (6) Fully connected layer

[0109] The 16 values ​​processed by the pooling layer are arranged in channel order and convolved with 4 sets of preset coefficients. By adding offset value adjustment, the weight value of each possible result is obtained. The one with the largest weight value is selected as the final classification result, that is, the fault type of this calculation is determined.

[0110] The original convolutional neural network usually requires tedious manual parameter adjustment, including the selection of learning rate, hierarchy and activation function. This process is not only time-consuming, but also requires rich professional experience to ensure the rationality of parameters and the effectiveness of the model. In order to simplify this process and improve model performance, the MTBO optimization CNN parameter process is adopted in this embodiment, and the MTBO (Mountain Team-based Optimization) algorithm is introduced to automatically optimize the parameter configuration of CNN. The specific steps are as follows:

[0111] 1. Define the problem

[0112] Set the parameter combination of CNN convolutional neural network to θ = {α, k, n k} to obtain the best parameter combination θ*, so that the objective function F(θ) on the validation set is maximized as the goal, and the objective function is set to a weighted combination of classification accuracy and training time:

[0113] F(θ)=Accuracy(θ)-λTrainming Time(θ);

[0114] Among them, α represents the learning rate, k represents the convolution kernel size, and n k represents the number of convolution kernels; Accuracy(θ) is the classification accuracy under the parameter combination θ, Training Time(θ) is the training time under the parameter combination θ, and λ is the time penalty coefficient;

[0115] 2. Initialization

[0116] Based on the mountaineering team optimization algorithm, each parameter in the hyperparameter is initialized and its range is defined as follows: α∈[α min ,α max ], k∈{3,5,7}, n k ∈[n k，min , n k，max ], where αmin and αmax are the minimum and maximum values ​​of parameter α respectively, and n k，min ,n k，max The parameters n are k Preset minimum and maximum values.

[0117] Define a mountaineering team, each mountaineering team includes multiple climbers, each climber corresponds to an initial parameter combination θ i ={α i , k i , n k，i}, where i is the subgroup index;

[0118] 3. Parameter evaluation

[0119] For each initial parameter combination θi, use it as the hyperparameter of the CNN model:

[0120] 1) Model construction and training: Build a CNN model and use the training data for training. 2) Calculate the corresponding accuracy Accuracy(θ) on the validation set, record the training time Training Time(θ), calculate the objective function F(θ), and evaluate and optimize the initial parameter combination θi based on the calculation results.

[0121] 4. Update and search

[0122] Based on the search strategy corresponding to MTBO, each sub-population is searched and updated according to the current fitness;

[0123] (1) Local search:

[0124] Compute the gradient for each initial parameter combination:

[0125]

[0126] in Represents the gradient of the current initial parameter combination; Represents the partial derivative of the objective function with respect to the learning rate α; is the partial derivative of the objective function with respect to the convolution kernel size k; The objective function is the number of convolution kernels n k The partial derivative of

[0127] The parameter combination after local search is:

[0128]

[0129] Among them, η is the step size parameter; θi is the current initial parameter combination, θ i-local is the parameter combination after local search;

[0130] (2) Cross-group learning:

[0131] The sub-groups share information and adjust their own parameters according to the best parameter combination of other teams. By adjusting their own parameters through information sharing, assuming that the best parameter combination for fitness in the tth generation is θ*, then:

[0132] θ i-cross =γ·θ * +(1-γ)·θ i-local ;

[0133] Among them, γ is the information reference factor; θ* is the parameter combination with the best fitness in the tth generation; θ i-cross is the parameter combination after cross-group learning.

[0134] 5. Integration and optimization

[0135] (1) Global Fusion:

[0136] Comprehensively combine the parameter combinations after cross-group learning and select the optimal parameter combination θ best :

[0137] θ best = argmaxF(θ icross );

[0138] Among them, argmax represents the independent variable that makes the function reach the maximum value, F(θ i-cross ) is the objective function to be maximized;

[0139] (2) Diversity maintenance:

[0140] Random mutation is introduced to maintain the diversity of the search and avoid early convergence. The updated new parameter combination θ(i,new) is expressed as follows:

[0141] θ i，new =θ i-cross +∈·randn();

[0142] Where, ∈ is the variation intensity; randn() is the standard normal random noise;

[0143] (3) Parameter update:

[0144] The updated new parameter combination θ(i,new) replaces the original θi as the starting point θ of the next iteration i ←θ i，new .

[0145] 6. Termination conditions

[0146] Determine whether the above optimization process meets the following termination conditions:

[0147] (1) Determine whether the maximum number of iterations Tmax has been reached;

[0148] (2) According to the comparison between the best fitness of the current t generation and the best fitness of the t-1 generation, it is judged whether the corresponding fitness value of the current optimal combination has been improved, that is:

[0149]

[0150] In the formula, δ is the convergence threshold, is the optimal parameter combination θ of the tth generation best The optimization result under is the fitness value;

[0151] 7. Iteration and Final Model

[0152] If the above termination conditions are met, the optimization process is terminated and the final optimal parameter combination configuration θ is determined. best And train the final model;

[0153] If the above termination condition is not met, return to the step of evaluating and optimizing the initial parameter combination θi, that is, step 3 to continue iterating.

[0154] Traditional fault diagnosis models often use manual parameter adjustment to optimize the model, which is time-consuming and requires professional knowledge. The MTBO-CNN model introduces the MTBO (Mountaineering Team-Based Optimization) algorithm, which uses the optimization algorithm to automatically search for the best parameter configuration without manual parameter adjustment, greatly improving the efficiency and accuracy of parameter optimization. The MTBO-CNN model uses convolutional neural network (CNN) as the basic model and has a strong feature extraction capability. CNN can automatically learn and extract important features from raw data, which plays a key role in fault identification and diagnosis. The MTBO-CNN model considers the influence of uncertainty and noise in the parameter optimization process, improving the robustness and stability of the model. In addition, by optimizing CNN parameters, the MTBO-CNN model can also improve the interpretability of the model, allowing engineers to better understand the fault diagnosis results and more accurately determine the cause of the fault. Compared with the use of CNN models or other traditional methods for fault diagnosis alone, the MTBO-CNN model has advantages in parameter optimization and feature extraction, so it can improve the accuracy and performance of fault diagnosis. By automatically optimizing CNN parameters, better results can be achieved in fault diagnosis tasks. The MTBO-CNN model fully considers the transferability of the model during the training process. By pre-training on a large-scale dataset and then fine-tuning on the target task, the model can better adapt to different circuit breaker fault diagnosis tasks. This can reduce the need for a large amount of labeled data and improve the training efficiency and generalization ability of the model.

[0155] A specific example of executing the fault detection process based on the above optimization steps is as follows:

[0156] (1) Data collection: In different fault scenarios, the sensor collects vibration signals to form a sample data set containing various fault levels, including the original vibration signals in four different states. Figure 10-13 shown.

[0157] (2) Using MTBO to optimize CNN: According to the mountaineering team optimization algorithm, the convolutional neural network learning rate, convolution kernel size and number of convolution kernels are optimized. The optimized parameters are shown in Table 1.

[0158] Table 1 CNN optimization parameter results

[0159]

[0160] (3) Model training and evaluation: Use the optimized parameter configuration to train the collected data and quantitatively evaluate the classification performance of the model under different fault scenarios to ensure that the network can accurately identify various mechanical faults of the circuit breaker. Use the test error as the radius label and the number of iterations as the angle label to draw the error iteration change curve as shown in the figure. Fig.14 As shown in the analysis chart, after 6 rounds of training, the training error dropped to 0.0008 and tended to be stable.

[0161] In order to further investigate the ability of MTBO-CNN model to identify circuit breaker faults, the confusion matrix is ​​introduced to quantitatively analyze the diagnosis results. The confusion matrix reflects in detail the number and type of misjudgments under different rolling bearing fault types, such as Fig.15 shown. Fig.15 Middle: The horizontal axis represents the predicted categories of different faults; the vertical axis represents the true labels of different faults; the numbers on the main diagonal of the matrix represent the number of samples correctly classified for each type of fault. The results show that for circuit breakers in four different states, 100% recognition accuracy can be achieved.

[0162] Embodiment 2

[0163] Based on the same concept, the present invention provides a circuit breaker fault detection optimization system based on convolutional neural network, comprising:

[0164] A data acquisition module, used for acquiring an original vibration signal generated when the circuit breaker is in operation;

[0165] A preprocessing module, used to preprocess the original vibration signal, and extract features from the preprocessed original vibration signal based on a convolutional neural network to obtain an extracted target feature data set;

[0166] MTBO-CNN model, used to build the MTBO-CNN model, using the CNN convolutional neural network as the basic model, and introducing the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model;

[0167] A model training module, used for dividing the target feature data set into a training set and a test set according to a preset ratio, and using the training set to train the MTBO-CNN model, and using the test set to test the MTBO-CNN model, to obtain a trained MTBO-CNN model;

[0168] The fault detection module is used to input the extracted target feature data into the trained MTBO-CNN model, and use the optimized network to classify the circuit breaker vibration signal to obtain the circuit breaker fault identification result.

[0169] It should be noted that the division of the modules in the embodiment of the device / system is only a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or physically separated. Moreover, these modules can be implemented in the form of software calling through processing elements; or in the form of hardware; or some units can be implemented in the form of software calling through processing elements, and some units can be implemented in the form of hardware.

[0170] The implementation principle of the above modules has been described in the above embodiments, so it will not be repeated here.

[0171] Embodiment 3

[0172] Based on the same concept, in some embodiments of the present application, an electronic device is also provided. The electronic device includes a memory and a processor, wherein the memory is used to store a processing program, and the processor executes the processing program according to instructions. When the processor executes the processing program, the circuit breaker fault detection optimization method based on a convolutional neural network in the above embodiment is implemented.

[0173] In some embodiments of the present application, a readable storage medium is further provided, which may be a non-volatile readable storage medium or a volatile readable storage medium. The readable storage medium stores instructions, and when the instructions are executed on a computer, an electronic device including the readable storage medium executes the aforementioned circuit breaker fault detection optimization method based on a convolutional neural network.

[0174] It is understandable that, for the aforementioned circuit breaker fault detection optimization method based on convolutional neural network, if it is implemented in the form of software function modules and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0175] Computer readable storage media may include data signals propagated in baseband or as part of a carrier wave, wherein readable program codes are carried. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program codes contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0176] The program code involved in executing the technical solution disclosed in this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Python, C++, etc., and also conventional procedural programming languages ​​such as C language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A circuit breaker fault detection optimization method based on convolutional neural network, characterized in that: include: Acquire the original vibration signal generated when the circuit breaker is in operation; Preprocessing the original vibration signal, and extracting features of the preprocessed original vibration signal based on a convolutional neural network to obtain an extracted target feature data set; Construct the MTBO-CNN model, using the CNN convolutional neural network as the basic model, and introduce the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model; Dividing the target feature data set into a training set and a test set according to a preset ratio, and using the training set to train the MTBO-CNN model, and using the test set to test the MTBO-CNN model, to obtain a trained MTBO-CNN model; The extracted target feature data is input into the trained MTBO-CNN model, and the optimized network is used to classify the circuit breaker vibration signal to obtain the circuit breaker fault identification result.

2. The circuit breaker fault detection optimization method based on convolutional neural network according to claim 1 is characterized in that: The obtaining of the original vibration signal generated when the circuit breaker is in operation further comprises: Construct a vibration signal acquisition system with a circuit breaker, set a vibration sensor on the circuit breaker, and start a working operation mode to simulate different mechanical fault types of the circuit breaker; Based on the vibration signal acquisition system, the signal acquisition device is activated to capture the analog vibration signal of the circuit breaker in the mechanical state, and the captured analog vibration signal is converted into a digital signal, which is collected and recorded according to a preset sampling rate, and the collected analog vibration signal is transmitted to the fault diagnosis system to perform circuit breaker fault detection operation; Among them, the vibration signal acquisition system includes a circuit breaker, a signal acquisition device and a fault diagnosis system connected in sequence, and the signal acquisition device is powered by an external power supply to ensure the normal operation of the acquisition work; the simulated vibration signal includes loose screws on the base of the circuit breaker, stuck iron core, insufficient lubrication and related fault signals of different degrees simulated by adjusting the corresponding mechanical parts on the circuit breaker.

3. The circuit breaker fault detection optimization method based on convolutional neural network according to claim 1 is characterized in that: In the MTBO-CNN model, a CNN convolutional neural network is used to perform signal processing and feature extraction in sequence according to the operation sequence of the neural network, which can efficiently realize multi-level information fusion. The CNN convolutional neural network includes an input layer, a one-dimensional convolutional layer, a two-dimensional convolutional layer, a correction layer, a maximum pooling layer and a full connection layer connected in sequence. The original vibration signal is received from the test source output as the input of the input layer, and is transmitted to the next network layer layer by layer to output the fault identification result.

4. The circuit breaker fault detection optimization method based on convolutional neural network according to claim 1, characterized in that: The introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model includes: The parameter combination of the CNN convolutional neural network is defined as θ = {α, k, n k }, to obtain the best parameter combination θ*, so that the objective function F(θ) on the validation set is maximized as the goal, and the objective function is set as a weighted combination of classification accuracy and training time F(θ)=Accuracy(θ)-λTraining Time(θ); Among them, α represents the learning rate, k represents the convolution kernel size, and n k represents the number of convolution kernels; Accuracy(θ) is the classification accuracy under the parameter combination θ, Training Time(θ) is the training time under the parameter combination θ, and λ is the time penalty coefficient; Based on the mountaineering team optimization algorithm, each parameter in the hyperparameter is initialized, and a mountaineering team is defined. Each mountaineering team includes multiple climbers, and each climber corresponds to an initial parameter combination θ i ={α i , k i , n k，i }, where i is the subgroup index; Take each initial parameter combination θi as the hyperparameter of the CNN model, train the CNN model, calculate the corresponding accuracy Accuracy(θ) on the validation set, record the training time Training Time(θ), calculate the objective function F(θ), and evaluate and optimize the initial parameter combination θi according to the calculation results.

5. The circuit breaker fault detection optimization method based on convolutional neural network according to claim 1, characterized in that: The introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model further includes: Based on the search strategy corresponding to MTBO, each sub-population is searched and updated according to the current fitness; If a local search is performed, the gradient is calculated for each initial parameter combination: in Represents the gradient of the current initial parameter combination; Represents the partial derivative of the objective function with respect to the learning rate α; is the partial derivative of the objective function with respect to the convolution kernel size k; The objective function is the number of convolution kernels n k The partial derivative of The parameter combination after local search is: Among them, η is the step size parameter; θi is the current initial parameter combination, θ i-local is the parameter combination after local search; If cross-group learning is performed, the parameters are adjusted through information sharing. Assuming that the best parameter combination for fitness in the tth generation is θ*, then: i i-cross =γ·θ * +(1-γ)·θ i-local ; Among them, γ is the information reference factor; θ* is the parameter combination with the best fitness in the tth generation; θ i-cross is the parameter combination after cross-group learning.

6. The circuit breaker fault detection optimization method based on convolutional neural network according to claim 1, characterized in that: The introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model further includes: Comprehensively combine the parameter combinations after cross-group learning and select the optimal parameter combination θ best : θ best =argmaxF(θ i-cross ); Among them, argmax represents the independent variable that makes the function reach the maximum value, F(θ i-cross ) is the objective function to be maximized; Introducing random mutations to maintain the diversity of the search, the updated new parameter combination θ(i,new) is expressed as follows: θ i,new =θ i-cross +∈randn(); Where, ∈ is the variation intensity; randn() is the standard normal random noise; The updated new parameter combination θ(i,new) replaces the original θi as the starting point for the next round of iteration.

7. The circuit breaker fault detection optimization method based on convolutional neural network according to claim 6 is characterized in that: The introduction of the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model further includes: Determine whether the above optimization process meets the following termination conditions: Determine whether the maximum number of iterations Tmax has been reached; and / or, According to the comparison between the best fitness of the current t generation and the best fitness of the t-1 generation, it is judged whether the corresponding fitness value of the current optimal combination has been improved, that is: In the formula, δ is the convergence threshold, is the optimal parameter combination θ of the tth generation best The optimization result under is the fitness value; If the above termination conditions are met, the optimization process is terminated and the final optimal parameter combination configuration θ is determined. best And train the final model; If the above termination conditions are not met, the process returns to the step of evaluating and optimizing the initial parameter combination θi and continues the iteration.

8. A circuit breaker fault detection optimization system based on convolutional neural network, characterized in that: include: A data acquisition module, used for acquiring an original vibration signal generated when the circuit breaker is in operation; A preprocessing module, used to preprocess the original vibration signal, and extract features from the preprocessed original vibration signal based on a convolutional neural network to obtain an extracted target feature data set; MTBO-CNN model, used to build the MTBO-CNN model, using the CNN convolutional neural network as the basic model, and introducing the MTBO algorithm to automatically search for the best parameter configuration, thereby optimizing the hyperparameters of the MTBO-CNN model; A model training module, used for dividing the target feature data set into a training set and a test set according to a preset ratio, and using the training set to train the MTBO-CNN model, and using the test set to test the MTBO-CNN model, to obtain a trained MTBO-CNN model; The fault detection module is used to input the extracted target feature data into the trained MTBO-CNN model, and use the optimized network to classify the circuit breaker vibration signal to obtain the circuit breaker fault identification result.

9. An electronic device, characterized in that: include: A memory, the memory being used to store a processing program; A processor, wherein when executing the processing program, the processor implements the circuit breaker fault detection optimization method based on convolutional neural network as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores a processing program, and when the processing program is executed by the processor, the circuit breaker fault detection optimization method based on convolutional neural network as described in any one of claims 1 to 7 is implemented.

Images