High-voltage circuit breaker fault detection method and device, electronic equipment and storage medium

By simulating the working conditions of the high-voltage circuit breaker, collecting and analyzing vibration signals, combining time-frequency diagrams and preset fault types, and training detection models, the problems of inaccurate and low reliability in the existing technology are solved, and high efficiency is achieved. Automatic fault identification.

CN119063976APending Publication Date: 2024-12-03DALI POWER SUPPLY BUREAU YUNNAN POWER GRID
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411002729.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-25
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Existing high-voltage circuit breaker fault detection methods are difficult to accurately identify different fault types, and are susceptible to noise interference, resulting in poor detection reliability.

Method used

By simulating the working faults of the spring actuator under different working conditions, vibration signals are collected and time-frequency diagrams are generated, and the preset fault type is associated, and the detection model is trained to achieve automatic fault detection.

Benefits of technology

It improves the reliability and efficiency of high-voltage circuit breaker fault detection, reduces the dependence on manual judgment, and can accurately identify different types of faults.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119063976A_ABST
    Figure CN119063976A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses a high-voltage circuit breaker fault detection method and device, electronic equipment and a storage medium, and the method comprises the steps: simulating the working faults of a spring operating mechanism under different working conditions, and collecting vibration signals corresponding to different working faults of the spring operating mechanism through a virtual instrument; generating a time-frequency diagram corresponding to the vibration signal; associating a preset fault type with the time-frequency graph to obtain a training set; and training a preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, and performing fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model. According to the scheme, the fault detection reliability of the high-voltage circuit breaker can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a high-voltage circuit breaker fault detection method, device, electronic equipment and storage medium. Background Art

[0002] Mechanical failure is the main type of failure of high-voltage circuit breakers. Early detection of defects and failures in the circuit breaker electrical control circuit and operating mechanism can not only avoid circuit breaker operation failures and accidents, but also provide the necessary basis for circuit breaker fault maintenance.

[0003] At present, when a high-voltage circuit breaker has a mechanical fault, the vibration signal contained in its mechanical characteristics will change accordingly. The circuit breaker's action signal is an impact signal, which is highly complex, instantaneous, irregular, and non-periodic and stable, and is susceptible to noise interference. In addition, due to the different equipment models, fault types, and acquisition schemes of high-voltage circuit breakers, it is difficult to fully obtain data on different fault types of different high-voltage short-circuit breakers in actual applications, resulting in the need to rely on manual judgment when performing subsequent high-voltage circuit breaker fault detection. It can be seen that the reliability of current high-voltage circuit breaker fault detection is poor. Summary of the invention

[0004] The embodiments of the present application provide a high-voltage circuit breaker fault detection system, method, electronic device and storage medium, which can detect different high-voltage circuit breaker faults and improve the reliability of high-voltage circuit breaker fault detection.

[0005] In a first aspect, an embodiment of the present application provides a high-voltage circuit breaker fault detection method, comprising:

[0006] Simulating working failures of the spring operating mechanism under different working conditions, and collecting vibration signals corresponding to different working failures of the spring operating mechanism through a virtual instrument;

[0007] Generating a time-frequency diagram corresponding to the vibration signal;

[0008] Associating the preset fault type with the time-frequency diagram to obtain a training set;

[0009] A preset detection model is trained based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

[0010] Optionally, in some embodiments of the present application, generating a time-frequency diagram corresponding to the vibration signal includes:

[0011] The vibration signal is subjected to wavelet transformation to obtain a time-frequency diagram corresponding to the vibration signal, and the format of the time-frequency diagram is converted into a preset format.

[0012] Optionally, in some embodiments of the present application, performing a wavelet transform on the vibration signal to obtain a time-frequency diagram corresponding to the vibration signal includes:

[0013] Get wavelet function and scaling function;

[0014] The vibration signal is subjected to wavelet transformation according to the wavelet function and the scaling function to obtain a time-frequency diagram corresponding to the vibration signal.

[0015] Optionally, in some embodiments of the present application, associating the preset fault type with the time-frequency graph to obtain a training set includes:

[0016] Determining the correlation between the preset fault type and the time-frequency diagram;

[0017] The fault types of the time-frequency graph are labeled according to the association relationship to obtain a training set.

[0018] Optionally, in some embodiments of the present application, the step of training a preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, and performing fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model, includes:

[0019] Get the preset detection model;

[0020] Determining positive samples and negative samples in the training set, and combining the positive samples and negative samples to obtain sample pairs;

[0021] inputting the positive samples of the sample pair into the first sub-network of the high-voltage circuit breaker fault detection model, and;

[0022] Inputting the negative sample of the sample pair into the second sub-network of the high-voltage circuit breaker fault detection model;

[0023] Training a preset detection model according to the positive sample features output by the first sub-network and the negative sample features output by the second sub-network to obtain a high-voltage circuit breaker fault detection model;

[0024] The high-voltage circuit breaker fault detection model is used to perform fault detection on the high-voltage circuit breaker.

[0025] Optionally, in some embodiments of the present application, the preset detection model is trained according to the positive sample features output by the first sub-network and the negative sample features output by the second sub-network to obtain a high-voltage circuit breaker fault detection model, including:

[0026] Detecting similarity between positive sample features output by the first sub-network and negative sample features output by the second sub-network;

[0027] The preset detection model is trained according to the detection results and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

[0028] Optionally, in some embodiments of the present application, the detecting the similarity between the positive sample features output by the first sub-network and the negative sample features output by the second sub-network includes:

[0029] Calculating a feature distance between positive sample features output by the first sub-network and negative sample features output by the second sub-network;

[0030] The method of training a preset detection model according to the detection results and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model includes: calculating the feature loss corresponding to the feature distance according to a preset contrast loss function, and training the preset detection model based on the feature loss and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

[0031] In a second aspect, an embodiment of the present application provides a high-voltage circuit breaker fault detection device, comprising:

[0032] A simulation module, used for simulating working failures of the spring operating mechanism under different working conditions, and collecting vibration signals corresponding to different working failures of the spring operating mechanism through a virtual instrument;

[0033] A generating module, used for generating a time-frequency diagram corresponding to the vibration signal;

[0034] An association module, used to associate a preset fault type with the time-frequency diagram to obtain a training set;

[0035] A training module is used to train a preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

[0036] Correspondingly, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the steps of any of the above methods when executing the program.

[0037] The present application also provides a storage medium, wherein the storage medium stores a computer program, and the computer program implements any of the above methods when the computer program is executed by a processor.

[0038] The present application provides a high-voltage circuit breaker fault detection method, device, electronic device and storage medium, which simulate the working fault of the spring operating mechanism under different working conditions, and collect the vibration signals corresponding to the different working faults of the spring operating mechanism through a virtual instrument, then generate a time-frequency diagram corresponding to the vibration signal, then associate the preset fault type with the time-frequency diagram to obtain a training set, and finally, train the preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model. In the scheme of high-voltage circuit breaker fault detection provided by the present application, the vibration signals corresponding to the different working faults of the spring operating mechanism are collected in advance through a virtual instrument, and the preset detection model is trained based on the vibration signal to obtain a high-voltage circuit breaker fault detection model. When performing high-voltage circuit breaker fault detection, the high-voltage short circuit fault detection model can be used to perform fault detection on different types of high-voltage circuit breakers without relying on manual judgment, which not only improves the reliability of high-voltage circuit breaker fault detection, but also improves the detection efficiency of high-voltage circuit breaker faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 It is a flow chart of a high-voltage circuit breaker fault detection method provided in an embodiment of the present application;

[0041] Figure 2 Schematic diagram of standing wave measurement distance in a high voltage circuit breaker fault detection method provided in an embodiment of the present application;

[0042] Figure 3 It is a structural schematic diagram of a high-voltage circuit breaker fault detection model in a high-voltage circuit breaker fault detection method provided in an embodiment of the present application;

[0043] Figure 4 It is a classification effect diagram for the accuracy of operating mechanism fault identification;

[0044] Figure 5 It is a structural schematic diagram of a high-voltage circuit breaker fault detection device provided in an embodiment of the present application;

[0045] Figure 6 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of systems and methods consistent with the examples detailed in the attached claims, or with some aspects of the present application.

[0047] It should be noted that, in this article, the terms "include", "comprises" or any other variants thereof are intended to cover descriptions such as non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element. In addition, components, features, and elements with the same name in different embodiments of the present application may have the same meaning or different meanings, and their specific meanings need to be determined by their explanation in the specific embodiment or further combined with the context of the specific embodiment.

[0048] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0049] In the subsequent description, the suffixes such as "module", "component" or "unit" used to represent elements are only used to facilitate the description of the present application, and have no specific meanings. Therefore, "module", "component" or "unit" can be used in a mixed manner.

[0050] Embodiments of the present application provide a high-voltage circuit breaker fault detection method, device, electronic device, and storage medium.

[0051] Among them, the high-voltage circuit breaker fault detection device can be deployed in a terminal or a server. The server can include an independently operated server or a distributed server, or a server cluster composed of multiple servers. The terminal can include a mobile phone, a tablet computer or a personal computer (PC, Personal Computer).

[0052] It should be noted that the order of description of the following embodiments is not intended to limit the priority order of the embodiments.

[0053] The present application provides a high-voltage circuit breaker fault detection method, comprising: simulating working faults of a spring operating mechanism under different working conditions, and collecting vibration signals corresponding to different working faults of the spring operating mechanism through a virtual instrument; generating a time-frequency graph corresponding to the vibration signal; associating a preset fault type with the time-frequency graph to obtain a training set; training a preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

[0054] See also Figure 1 , Figure 1 : is a flow chart of a high-voltage circuit breaker fault detection method provided by the present application, and the high-voltage circuit breaker fault detection method specifically includes the following process:

[0055] 101. Simulate the working failure of the spring operating mechanism under different working conditions, and collect the vibration signals corresponding to the different working failures of the spring operating mechanism through virtual instruments.

[0056] Vibration signals are oscillations or fluctuations generated by a mechanical system or structure during motion, which can be measured by various sensors such as accelerometers, velocimeters or displacement meters. Vibration signal analysis is an important technology in the engineering field, which is used to monitor and diagnose the operating status of the machine and identify faults or abnormal behaviors.

[0057] For example, specifically, experiments can be designed to simulate various fault conditions that may be encountered by the spring operating mechanism. These faults may include: core jamming, opening spring stress relaxation, transmission mechanism jamming, oil buffer abnormality, cam wear, etc. Then, determine different operating conditions, such as different loads, different operating frequencies, different environmental conditions (such as temperature, humidity), etc., to simulate various conditions in actual operation. Use virtual instrument software (such as LabVIEW, MATLAB, etc.) to configure the data acquisition system. Virtual instruments can provide flexible signal processing and data acquisition functions. Install vibration sensors, such as accelerometers or velocity sensors, at key parts of the spring operating mechanism to capture vibration signals generated by mechanical movement. Start the virtual instrument and start collecting vibration signals under simulated faults and different working conditions. Ensure the stability of the acquisition process and the accuracy of the data. Store the collected vibration signal data in a computer, usually in a digital format, such as a CSV (Comma-Separated Values) file or a MAT (a data storage format generated by MATLAB software) file.

[0058] 102. Generate a time-frequency diagram corresponding to the vibration signal.

[0059] A time-frequency diagram is a visualization tool used to show the characteristics of a signal in two dimensions: time and frequency. This graphical representation is particularly useful for analyzing non-stationary signals, that is, signals whose statistical properties change over time. A time-frequency diagram shows both the time domain waveform and the frequency domain characteristics of a signal, which helps to understand the time variation and frequency components of a signal. It is particularly suitable for analyzing non-stationary signals, such as vibration signals. Time-frequency diagrams are often displayed in the form of heat maps, where the depth of color represents the strength or energy of the signal at a specific time and frequency.

[0060] For example, specifically, the vibration signal can be preprocessed, including filtering to remove noise, detrending to eliminate DC components, etc., and then a wavelet function suitable for the signal characteristics is determined for time-frequency analysis. Commonly used wavelet functions include Morlet wavelet, Haar wavelet, etc. Apply wavelet transform to the preprocessed signal to obtain the time-frequency representation of the signal. Wavelet transform can reveal the characteristics of the signal at different times and frequencies. Calculate the wavelet transform coefficients, which represent the intensity of the signal at different scales (frequency) and positions (time). Map the amplitude or energy of the wavelet transform coefficients to the time-frequency plane, with the horizontal axis representing time, the vertical axis representing frequency, and the color depth representing the intensity of the transform coefficient.

[0061] Optionally, in some embodiments of the present application, the step of “generating a time-frequency diagram corresponding to the vibration signal” may specifically include:

[0062] The vibration signal is subjected to wavelet transformation to obtain a time-frequency diagram corresponding to the vibration signal, and the format of the time-frequency diagram is converted into a preset format.

[0063] For example, specifically, a suitable wavelet function is determined according to the characteristics of the signal for time-frequency analysis. Commonly used wavelet functions include Morlet wavelet, Haar wavelet, etc. The selected wavelet function is used to perform continuous wavelet transform (CWT) or discrete wavelet transform (DWT) on the preprocessed vibration signal. The result of the wavelet transform is used to generate a time-frequency representation of the signal, usually by calculating the amplitude or energy of the wavelet coefficients. The time-frequency representation is plotted as a graph, and a heat map is usually used to show the intensity of the wavelet coefficients at different times and frequencies.

[0064] Optionally, in some embodiments of the present application, the step of “performing a wavelet transform on the vibration signal to obtain a time-frequency diagram corresponding to the vibration signal” may specifically include:

[0065] Get wavelet function and scaling function;

[0066] The vibration signal is subjected to wavelet transform according to the wavelet function and the scaling function to obtain the time-frequency diagram corresponding to the vibration signal.

[0067] Determine which wavelet function to use. The choice of wavelet function depends on the characteristics of the signal and the needs of the analysis. Common wavelet functions include Morlet, Haar, Daubechies, etc. In continuous wavelet transform (CWT), scale and translation parameters are the key to the transformation. Scale parameters are usually related to frequency, while translation parameters are aligned with time. For the selected wavelet function, calculate its scale function. The scale function defines the shape and size of the wavelet at different scales. Transform the vibration signal using the selected wavelet function and scale function. Wavelet transform usually involves convolving the signal with the wavelet function at different scales and translation parameters. The result of wavelet transform is a time-frequency representation that contains the energy or amplitude information of the signal at different time and scales (frequency). Plot the time-frequency representation into a graph, usually using a heat map. In this graph, the depth of color represents the amplitude or energy of the wavelet transform coefficients.

[0068] In some embodiments of the present application, the collected vibration signal is analyzed by Morlet wavelet transform. Morlet wavelet transform can realize the analysis of different frequency components at different scales, reveal the frequency distribution of vibration signals at different scales, help analyze the characteristics of the signal in different frequency ranges, and enable the model to better extract the essential characteristics of the correlation between the change of the vibration signal of the spring operating mechanism and the mechanical failure of the high-voltage circuit breaker. By converting the signal into a time-frequency diagram, the characteristic representation of the signal can be learned more effectively, thereby improving the performance and generalization ability of the model in subsequent tasks. The formula of Morlet wavelet transform:

[0069]

[0070] Where: f(t) is the original signal, expressed as a function of time t; a is the scale variable, which determines the degree of expansion and contraction of the wavelet function, that is, the width or resolution of the wavelet. The smaller the scale, the wider the wavelet and the lower the resolution; the larger the scale, the narrower the wavelet and the higher the resolution; b is the translation factor, which determines the position of the wavelet function on the time axis; is the time variable; ω0 is the center frequency, which is the frequency parameter of the wavelet function and determines the frequency characteristics of the wavelet; i is the imaginary unit, which satisfies i 2 =-1, is a complex exponential function used to describe the phase change frequency modulation part of the wavelet function; is a Gaussian function, also known as the exponential form of the normal distribution function. This Gaussian function is usually used in the wavelet function in wavelet transform to provide localized characteristics about time. In wavelet transform, the wavelet function usually needs to be localized in both time and frequency, and the Gaussian function provides a method to achieve this localization in time. The time-frequency domain data obtained by wavelet transform of a one-dimensional time series signal is as follows: Figure 2 shown.

[0071] 103. Associate the preset fault type with the time-frequency diagram to obtain a training set.

[0072] For example, specifically, determine and define the possible fault types. These fault types should be pre-defined by domain knowledge or experts. For each fault type, assign a label to each time-frequency graph or extracted feature, i.e., the corresponding fault type. Combine the extracted features and the corresponding fault type labels to construct a dataset for training.

[0073] Optionally, in some embodiments of the present application, the step of “associating a preset fault type with a time-frequency graph to obtain a training set” may specifically include:

[0074] Determine the correlation between the preset fault type and the time-frequency diagram;

[0075] The fault types of the time-frequency graph are labeled according to the association relationship to obtain a training set.

[0076] For example, based on the results of the time-frequency graph analysis, establish the correlation between the fault type and the time-frequency features. Use the established correlation to label each time-frequency graph with the fault type. Convert the time-frequency graph labeled with the fault type into a data format that can be used for machine learning model training. This usually involves extracting feature vectors and assigning labels. Preprocess the training set data, including normalization and standardization, to improve the effect of model training.

[0077] 104. Train a preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

[0078] A detection model refers to a trained machine learning or deep learning model that can identify and classify different types of faults. For example, depending on the complexity of fault detection and the characteristics of the data, determine a suitable model architecture, which can be a traditional machine learning model such as random forest, SVM, decision tree, etc., or a more complex deep learning model such as convolutional neural network (CNN), recurrent neural network (RNN) or deep belief network (DBN). Preprocess the training set data, including feature scaling, normalization, standardization, etc., to eliminate the influence of different dimensions and improve the efficiency of model training. The training set is further divided into training data and validation data for model hyperparameter tuning and avoiding overfitting. According to the type and task of the model (such as classification or regression), choose the appropriate loss function and optimizer. For deep learning models, use appropriate compilation steps to configure the model training process. Train the model using the training data. During the training process, monitor the performance of the model on the validation set to detect overfitting or underfitting in a timely manner. Evaluate the performance of the model on independent test data, using indicators such as accuracy, recall, and F1 score. Based on the performance of the model on the validation data, adjust the model's hyperparameters, such as learning rate, batch size, number of layers, number of neurons, etc. Apply regularization techniques (such as L1, L2 regularization, Dropout) to prevent overfitting. Use techniques such as cross-validation to ensure the stability and generalization ability of the model. After training is completed, save the model's parameters for subsequent deployment and use.

[0079] Optionally, in some embodiments of the present application, the step of “training a preset detection model based on a training set to obtain a high-voltage circuit breaker fault detection model, and performing fault detection on a high-voltage circuit breaker through the high-voltage circuit breaker fault detection model” may specifically include:

[0080] Get the preset detection model;

[0081] Determine positive samples and negative samples in the training set, and combine the positive samples and negative samples to obtain sample pairs;

[0082] inputting positive samples of the sample pairs into the first sub-network of the high-voltage circuit breaker fault detection model, and;

[0083] Inputting the negative sample of the sample pair into the second sub-network of the high-voltage circuit breaker fault detection model;

[0084] The preset detection model is trained according to the positive sample features output by the first sub-network and the negative sample features output by the second sub-network to obtain a high-voltage circuit breaker fault detection model;

[0085] The high voltage circuit breaker fault detection model is used to detect the high voltage circuit breaker fault.

[0086] In an embodiment of the present application, the detection model may be a twin network (Siamese network), which is composed of two identical sub-networks that share the same weights. The Siamese network receives two sets of input data, which may be images, vectors, or other types of data. Each sub-network independently extracts features of its input data. The Siamese network learns how to measure the similarity of data pairs by comparing the feature vectors output by the two sub-networks. Among them, the Siamese network uses a specific loss function (such as contrastive loss Contrastive Loss) to train the network so that similar data pairs have closer feature representations, while dissimilar data pairs have feature representations farther away. The weight sharing between the two sub-networks means that they use the same set of parameters to learn features. Siamese networks are widely used in tasks such as image retrieval, signature verification, and anomaly detection. During the training process, the network parameters are adjusted by optimizing the loss function to improve the accuracy of the similarity measurement. The trained Siamese network can be generalized to new, unseen data for effective similarity measurement.

[0087] Positive samples: refers to sample pairs that belong to the same category or have high similarity. For example, in fault detection, positive samples can be two different instances of the same fault type. Negative samples: refers to sample pairs that belong to different categories or have low similarity. In fault detection, negative samples can be instances of different fault types.

[0088] Combine the selected positive and negative samples to form a set of sample pairs for training. These sample pairs will be used to train the model to learn how to distinguish between similar and dissimilar samples.

[0089] For example, specifically, the Siamese network receives two sets of inputs, usually a pair of samples, which can be similar (positive samples) or dissimilar (negative sample pairs). Each subnetwork (i.e., the first subnetwork and the second subnetwork) independently extracts features of its input samples. By comparing the features output by the first subnetwork and the second subnetwork, the similarities and differences between the samples are learned. The network is trained using a contrastive loss function (such as Triplet Loss or Contrastive Loss) so that the feature representations of similar samples are closer and the feature representations of dissimilar samples are farther apart. The parameters of the two subnetworks are shared, which means that the learned feature representations are valid for both inputs. During the training process, both positive and negative sample pairs are optimized to learn the ability to distinguish between different categories. After the training is completed, the Siamese network can be used for fault detection of high-voltage circuit breakers, and the trained Siamese network can generalize to new, unseen abnormal data.

[0090] Optionally, in some embodiments of the present application, the step of “training a preset detection model according to the positive sample features output by the first sub-network and the negative sample features output by the second sub-network to obtain a high-voltage circuit breaker fault detection model” may specifically include:

[0091] Detecting the similarity between the positive sample features output by the first sub-network and the negative sample features output by the second sub-network;

[0092] The preset detection model is trained according to the detection results and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

[0093] Select positive sample pairs (same fault type) and negative sample pairs (different fault types) from the training set. Input the positive sample to the first subnetwork of the Siamese network to extract its features; input the negative sample to the second subnetwork of the Siamese network to extract its features. Calculate the similarity between the positive sample features output by the first subnetwork and the negative sample features output by the second subnetwork. This can be achieved by calculating the distance between the feature vectors (such as Euclidean distance or cosine similarity). Use a contrastive loss function (such as Triplet Loss or Contrastive Loss) to measure the similarity difference between the positive sample pairs and the negative sample pairs. The goal is to make the features of the positive sample pairs closer and the features of the negative sample pairs farther apart. According to the results of the loss function, the network weights are updated through the back-propagation algorithm so that the network can better distinguish between the positive and negative sample pairs in the next forward propagation. Repeat the above process until the network achieves satisfactory performance in distinguishing between positive and negative sample pairs. This usually involves multiple iterations of training. Use a validation set or a test set to evaluate the performance of the model to ensure that the model not only performs well on the training set, but also generalizes on unseen data. Adjust the model's hyperparameters, such as learning rate, batch size, network depth, etc., based on the evaluation results to optimize the model performance. Apply regularization techniques (such as Dropout, L1 / L2 regularization) to prevent overfitting and improve the generalization ability of the model. After training, the obtained Siamese network model can be used for high-voltage circuit breaker fault detection tasks. Deploy the trained model to the actual fault detection system to perform real-time or batch fault detection on new vibration signal data.

[0094] Optionally, in some embodiments of the present application, the step of “detecting the similarity between the positive sample features output by the first sub-network and the negative sample features output by the second sub-network” may specifically include:

[0095] Calculate the feature distance between the positive sample features output by the first sub-network and the negative sample features output by the second sub-network.

[0096] Feature distance is an indicator that measures the similarity between two feature vectors. In Siamese networks, feature distance is used to quantify the similarity or difference between input samples processed by the network. Examples of features include Euclidean distance, Hamming distance, or Mahalanobis distance, which can be selected based on actual conditions.

[0097] Optionally, in some embodiments of the present application, the step of "training a preset detection model according to the detection results and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model" may specifically include: calculating the feature loss corresponding to the feature distance according to a preset contrast loss function, and training the preset detection model based on the feature loss and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

[0098] For example, for each pair of positive and negative samples, the distance between their feature vectors is calculated, and methods such as Euclidean distance and cosine similarity can be used. The loss value is calculated using the contrast loss function and the calculated feature distance. The goal of the loss function is to make the distance of the positive sample pair small and the distance of the negative sample pair large. According to the calculated loss value, the weight of the network is updated through the back propagation algorithm. The purpose is to minimize the loss function so that the model can better distinguish between positive and negative sample pairs. Repeat the above process using the data of the training set until the model achieves satisfactory performance in distinguishing between positive and negative sample pairs.

[0099] For further information, see Figure 3 , by introducing the design of residual block, the gradient disappearance problem is effectively solved, a deeper network structure is constructed, and the performance of the network is improved. The core idea of ​​residual learning is to let the network learn the residual mapping between input and output, that is, H(x) = F(x) + x, where H(x) is the desired mapping, F(x) is the residual mapping, and x is the input. In this way, the network can learn the residual of the input data instead of directly learning the mapping H(x). The advantage of this is that when the residual mapping is close to 0, the network layer can "skip" some operations, which is equivalent to creating an identity mapping, which allows the network to be trained smoothly even in very deep cases.

[0100] The basic building block of ResNet is the residual block. Each residual block contains the following key parts:

[0101] Convolutional Layer: A regular convolutional layer for feature extraction. Activation Function: Usually a ReLU (Rectified Linear Unit) activation function. Identity Shortcut Connection: The input is directly connected to the following layer through an identity operation (such as a 1x1 convolution) so that it can be added to the output of the convolutional layer. Batch Normalization: Batch Normalization is used after each convolutional layer to help speed up the training process and improve the generalization ability of the model.

[0102] The Basic Block contains two convolutional layers, each followed by a batch normalization and ReLU activation function. Bottleneck Block: In order to reduce the number of parameters and computation, three convolutional layers are introduced, where the first and last convolutional layers have fewer channels and the middle convolutional layers have more channels. This design is called "1x1 convolution".

[0103] The hierarchical structure of ResNet is composed of multiple residual blocks stacked together, each of which is usually followed by an average pooling layer and a fully connected layer for classification tasks. ResNet has many variants, such as ResNet-18, ResNet-34, ResNet-50, ResNet-101, and ResNet-152, where the numbers represent the number of layers in the network. ResNet-50 and above models are built using bottleneck residual blocks.

[0104] The entire network provided by this application consists of convolutional layers, batch normalization layers, residual blocks and fully connected layers. It contains 33 convolutional layers and 1 fully connected layer. Each residual block contains several convolutional layers and jump connections, and the input signal is added to the output signal through the residual connection.

[0105] The feature difference between positive and negative samples is measured by the contrastive loss function, which aims to shorten the distance between positive samples and increase the distance between negative samples to learn more discriminative feature representations. The feature representation trained by contrastive learning is then input into a multilayer perceptron (MLP) for classification. The nonlinear transformation of the input features is achieved through several MLPs (fully connected layers) and ReLU activation functions to obtain the transformation vectors zi and zj. The contrastive loss is calculated using the one-dimensional vector obtained by the MLP, and the encoder model parameters are optimized through the back-propagation algorithm.

[0106] These subnetworks share weights. Each subnetwork maps the input data to a feature space and then calculates the distance or similarity between these features. By minimizing the distance between similar samples and maximizing the distance between different samples, the network can learn the intrinsic representation of the data. It is a neural network architecture specifically used to compare the similarity or distance between two input data (such as images, text, etc.). The core of its design lies in weight sharing, and the two input data are processed by exactly the same network structure. The number of samples in each category in the training set is significantly different, and the Siamese network can also effectively learn and accurately compare the similarity between two input data.

[0107] Since the output of the Siamese network is a measure of similarity between two input data, it can be easily combined with other machine learning algorithms to build an integrated learning model to further improve model performance. The Siamese network is able to deeply learn the semantic similarity of the input data, not just simple feature matching. This makes it perform well in dealing with complex problems such as image recognition and natural language processing. The structure and training method of the Siamese network can be flexibly adjusted to adapt to different tasks and data sets. The performance of the model can be further optimized by adjusting parameters such as the number of network layers, the number of neurons, or the optimization algorithm. In addition, the Siamese network can also be combined with other deep learning models, such as convolutional neural networks (CNN) or recurrent neural networks (RNN), to further improve its scalability and adaptability.

[0108] Figure 4 This is the classification effect of the fault recognition accuracy of the actuator after the deep network joint contrast learning. The results show that the recognition accuracy is as high as 97% after the deep network joint contrast learning: Through contrast learning, the model can learn the intrinsic structure and relationship of the data, thereby generating a more discriminative feature representation. This representation not only helps to improve the performance of downstream tasks, but also plays a huge role in scenarios where data is scarce or the annotation cost is high. Joint contrast learning enables the model to learn more robust and universal features by comparing between different data sets or views. This helps the model maintain good performance when facing new and unseen data, thereby improving its generalization ability.

[0109] Traditional supervised learning requires a large amount of labeled data to train the model, while contrastive learning can be performed under unsupervised or weak supervision. This means that the model does not need to rely on a large amount of labeled data during training, thereby reducing the cost of data collection and labeling. Joint contrastive learning usually uses multi-view or multi-modal data for training, which allows the model to learn the intrinsic structure of the data from multiple angles and levels. This not only helps to improve the training efficiency of the model, but also improves the performance of the model to a certain extent. This makes it particularly suitable for handling such complex scenarios. Through contrastive learning, a large amount of unlabeled data can be used to pre-train the model, and then fine-tuned on a small amount of labeled data, thereby achieving efficient model training and application.

[0110] As can be seen from the above, the embodiment of the present application provides a high-voltage circuit breaker fault detection method, after simulating the working fault of the spring operating mechanism under different working conditions, and collecting the vibration signals corresponding to the different working faults of the spring operating mechanism through the virtual instrument, a time-frequency diagram corresponding to the vibration signal is generated, and then the preset fault type is associated with the time-frequency diagram to obtain a training set, and finally, the preset detection model is trained based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model. In the scheme of high-voltage circuit breaker fault detection provided by the present application, the vibration signals corresponding to the different working faults of the spring operating mechanism are collected in advance through the virtual instrument, and the preset detection model is trained based on the vibration signal to obtain the high-voltage circuit breaker fault detection model. When performing high-voltage circuit breaker fault detection, the high-voltage short circuit fault detection model can be used to perform fault detection on different types of high-voltage circuit breakers, without relying on manual judgment, which not only improves the reliability of high-voltage circuit breaker fault detection, but also improves the detection efficiency of high-voltage circuit breaker faults.

[0111] In order to better implement the high-voltage circuit breaker fault detection method of the embodiment of the present application, the embodiment of the present application also provides a high-voltage circuit breaker fault detection device. The meanings of the terms are the same as those in the above-mentioned high-voltage circuit breaker fault detection system, and the specific implementation details can refer to the description in the system embodiment.

[0112] See also Figure 5 , Figure 5 A schematic diagram of the structure of a high-voltage circuit breaker fault detection device provided in an embodiment of the present application, wherein the high-voltage circuit breaker fault detection device may specifically include a simulation module 201, a generation module 202, an association module 203 and a training module 204, which may be specifically as follows:

[0113] The simulation module 201 is used to simulate the working failures of the spring operating mechanism under different working conditions, and collect vibration signals corresponding to different working failures of the spring operating mechanism through a virtual instrument.

[0114] The generating module 202 is used to generate a time-frequency diagram corresponding to the vibration signal.

[0115] Optionally, in some embodiments of the present application, the generating module 202 may specifically include:

[0116] A transform unit, used for performing wavelet transform on the vibration signal to obtain a time-frequency diagram corresponding to the vibration signal;

[0117] The conversion unit is used to convert the format of the time-frequency diagram into a preset format.

[0118] Optionally, in some embodiments of the present application, the transformation unit can be specifically used to: obtain a wavelet function and a scaling function; perform a wavelet transform on the vibration signal according to the wavelet function and the scaling function to obtain a time-frequency diagram corresponding to the vibration signal.

[0119] The association module 203 is used to associate the preset fault type with the time-frequency diagram to obtain a training set.

[0120] Optionally, in some embodiments of the present application, the association module 203 may be specifically used to: determine the association relationship between a preset fault type and the time-frequency diagram; and mark the fault type of the time-frequency diagram according to the association relationship to obtain a training set.

[0121] The training module 204 is used to train a preset detection model based on a training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

[0122] Optionally, in some embodiments of the present application, the training module 204 may specifically include:

[0123] An acquisition unit, used for acquiring a preset detection model;

[0124] A determination unit, used to determine positive samples and negative samples in the training set, and combine the positive samples and the negative samples to obtain sample pairs;

[0125] An input unit, used to input the positive sample of the sample pair into the first sub-network of the high-voltage circuit breaker fault detection model, and to input the negative sample of the sample pair into the second sub-network of the high-voltage circuit breaker fault detection model;

[0126] A training unit, used to train a preset detection model according to the positive sample features output by the first sub-network and the negative sample features output by the second sub-network to obtain a high-voltage circuit breaker fault detection model;

[0127] The detection unit is used to perform fault detection on the high-voltage circuit breaker through a high-voltage circuit breaker fault detection model.

[0128] Optionally, in some embodiments of the present application, the training unit can be specifically used to: detect the similarity of the positive sample features output by the first sub-network and the negative sample features output by the second sub-network; train the preset detection model according to the detection results and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

[0129] Optionally, in some embodiments of the present application, the training unit can be specifically used to: calculate the feature distance between the positive sample features output by the first sub-network and the negative sample features output by the second sub-network; calculate the feature loss corresponding to the feature distance according to a preset contrast loss function, and train the preset detection model based on the feature loss and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

[0130] The embodiment of the present application provides a high-voltage circuit breaker fault detection device, wherein the simulation module 201 simulates the working fault of the spring operating mechanism under different working conditions, and after collecting the vibration signals corresponding to the different working faults of the spring operating mechanism through the virtual instrument, the generation module 202 generates the time-frequency diagram corresponding to the vibration signal, and then the association module 203 associates the preset fault type with the time-frequency diagram to obtain a training set, and finally, the training module 204 trains the preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model. In the scheme of high-voltage circuit breaker fault detection provided by the present application, the vibration signals corresponding to the different working faults of the spring operating mechanism are collected in advance through the virtual instrument, and the preset detection model is trained based on the vibration signal to obtain the high-voltage circuit breaker fault detection model. When performing high-voltage circuit breaker fault detection, the high-voltage short circuit fault detection model can be used to perform fault detection on different types of high-voltage circuit breakers, without relying on manual judgment, which not only improves the reliability of high-voltage circuit breaker fault detection, but also improves the detection efficiency of high-voltage circuit breaker faults.

[0131] In addition, the present application also provides an electronic device, such as Figure 6 As shown, it shows a schematic diagram of the structure of the electronic device involved in the embodiment of the present application, specifically:

[0132] The electronic device may include components such as a processor 301 with one or more processing cores, a memory 302 with one or more computer-readable storage media, a power supply 303, and an input unit 304. Those skilled in the art will appreciate that Figure 6 The electronic device structure shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0133] The processor 301 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. By running or executing software programs and / or modules stored in the memory 302, and calling data stored in the memory 302, it executes various functions of the electronic device and processes data, thereby monitoring the electronic device as a whole. Optionally, the processor 301 may include one or more processing cores; preferably, the processor 301 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 301.

[0134] The memory 302 can be used to store software programs and modules. The processor 301 executes various functional applications and a high-voltage circuit breaker fault detection method by running the software programs and modules stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 302 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. Accordingly, the memory 302 may also include a memory controller to provide the processor 301 with access to the memory 302.

[0135] The electronic device also includes a power supply 303 for supplying power to each component. Preferably, the power supply 303 can be logically connected to the processor 301 through a power management system, so that the power management system can manage charging, discharging, power consumption and other functions. The power supply 303 can also include one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, power status indicators and other arbitrary components.

[0136] The electronic device may further include an input unit 304, which may be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.

[0137] Although not shown, the electronic device may further include a display unit, etc., which will not be described in detail herein. Specifically in this embodiment, the processor 301 in the electronic device will load the executable files corresponding to the processes of one or more application programs into the memory 302 according to the following instructions, and the processor 301 will run the application programs stored in the memory 302, thereby realizing various functions, as follows:

[0138] The working failures of the spring operating mechanism under different working conditions are simulated, and the vibration signals corresponding to the different working failures of the spring operating mechanism are collected through virtual instruments; the time-frequency diagram corresponding to the vibration signal is generated; the preset fault type is associated with the time-frequency diagram to obtain a training set; the preset detection model is trained based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

[0139] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0140] The embodiment of the present application simulates the working failure of the spring operating mechanism under different working conditions, and after collecting the vibration signals corresponding to the different working failures of the spring operating mechanism through the virtual instrument, generates a time-frequency diagram corresponding to the vibration signal, and then associates the preset fault type with the time-frequency diagram to obtain a training set. Finally, the preset detection model is trained based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model. In the scheme of high-voltage circuit breaker fault detection provided by the present application, the vibration signals corresponding to the different working failures of the spring operating mechanism are collected in advance through the virtual instrument, and the preset detection model is trained based on the vibration signal to obtain the high-voltage circuit breaker fault detection model. When performing high-voltage circuit breaker fault detection, the high-voltage short circuit fault detection model can be used to perform fault detection on different types of high-voltage circuit breakers without relying on manual judgment, which not only improves the reliability of high-voltage circuit breaker fault detection, but also improves the detection efficiency of high-voltage circuit breaker faults.

[0141] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0142] To this end, an embodiment of the present application provides a storage medium in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the steps in any high-voltage circuit breaker fault detection method provided in the embodiment of the present application. For example, the instructions can execute the following steps:

[0143] The working failures of the spring operating mechanism under different working conditions are simulated, and the vibration signals corresponding to the different working failures of the spring operating mechanism are collected through virtual instruments; the time-frequency diagram corresponding to the vibration signal is generated; the preset fault type is associated with the time-frequency diagram to obtain a training set; the preset detection model is trained based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

[0144] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0145] The storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0146] Since the instructions stored in the storage medium can execute the steps in any high-voltage circuit breaker fault detection method provided in the embodiments of the present application, the beneficial effects that can be achieved by any high-voltage circuit breaker fault detection method provided in the embodiments of the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0147] The above is a detailed introduction to a high-voltage circuit breaker fault detection method, device, electronic device and storage medium provided in the embodiments of the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A high voltage circuit breaker fault detection method, It is characterized in that include: Simulating working failures of the spring operating mechanism under different working conditions, and collecting vibration signals corresponding to different working failures of the spring operating mechanism through a virtual instrument; Generating a time-frequency diagram corresponding to the vibration signal; Associating the preset fault type with the time-frequency diagram to obtain a training set; A preset detection model is trained based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

2. The high voltage circuit breaker fault detection method according to claim 1, It is characterized in that The generating a time-frequency diagram corresponding to the vibration signal includes: The vibration signal is subjected to wavelet transformation to obtain a time-frequency diagram corresponding to the vibration signal, and the format of the time-frequency diagram is converted into a preset format.

3. The high voltage circuit breaker fault detection method according to claim 2, It is characterized in that The step of performing wavelet transform on the vibration signal to obtain a time-frequency diagram corresponding to the vibration signal includes: Get wavelet function and scaling function; The vibration signal is subjected to wavelet transformation according to the wavelet function and the scaling function to obtain a time-frequency diagram corresponding to the vibration signal.

4. The high voltage circuit breaker fault detection method according to claim 1, It is characterized in that The step of associating the preset fault type with the time-frequency diagram to obtain a training set includes: Determining the correlation between the preset fault type and the time-frequency diagram; The fault types of the time-frequency graph are labeled according to the association relationship to obtain a training set.

5. The high voltage circuit breaker fault detection method according to claim 1, It is characterized in that The method of training a preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, and performing fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model, comprises: Get the preset detection model; Determining positive samples and negative samples in the training set, and combining the positive samples and negative samples to obtain sample pairs; inputting the positive samples of the sample pair into the first sub-network of the high-voltage circuit breaker fault detection model, and; Inputting the negative sample of the sample pair into the second sub-network of the high-voltage circuit breaker fault detection model; Training a preset detection model according to the positive sample features output by the first sub-network and the negative sample features output by the second sub-network to obtain a high-voltage circuit breaker fault detection model; The high-voltage circuit breaker fault detection model is used to perform fault detection on the high-voltage circuit breaker.

6. The high voltage circuit breaker fault detection method according to claim 5, It is characterized in that The method of training a preset detection model according to the positive sample features output by the first sub-network and the negative sample features output by the second sub-network to obtain a high-voltage circuit breaker fault detection model includes: Detecting similarity between positive sample features output by the first sub-network and negative sample features output by the second sub-network; The preset detection model is trained according to the detection results and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

7. The high voltage circuit breaker fault detection method according to claim 6, It is characterized in that The detecting the similarity between the positive sample features output by the first sub-network and the negative sample features output by the second sub-network includes: Calculating a feature distance between a positive sample feature output by the first sub-network and a negative sample feature output by the second sub-network; The method of training a preset detection model according to the detection results and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model includes: calculating the feature loss corresponding to the feature distance according to a preset contrast loss function, and training the preset detection model based on the feature loss and the labeled data of the training set to obtain a high-voltage circuit breaker fault detection model.

8. A high voltage circuit breaker fault detection device, It is characterized in that include: A simulation module, used for simulating working failures of the spring operating mechanism under different working conditions, and collecting vibration signals corresponding to different working failures of the spring operating mechanism through a virtual instrument; A generating module, used for generating a time-frequency diagram corresponding to the vibration signal; An association module, used to associate a preset fault type with the time-frequency diagram to obtain a training set; A training module is used to train a preset detection model based on the training set to obtain a high-voltage circuit breaker fault detection model, so as to perform fault detection on the high-voltage circuit breaker through the high-voltage circuit breaker fault detection model.

9. An electronic device, It is characterized in that include: A memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program according to the steps of the high-voltage circuit breaker fault detection method according to any one of claims 1 to 7.

10. A storage medium, It is characterized in that A computer program is stored which can be loaded by a processor and execute the high-voltage circuit breaker fault detection method according to any one of claims 1 to 7.

Citation Information

Cited By

  • Circuit breaker spring fatigue early warning diagnosis method and device and electronic equipment

    CN121207517A