Fault detection method and device for gas insulated switchgear, and electronic equipment
By converting the vibration signals of gas-insulated switch equipment into two-dimensional maps and applying a lightweight convolutional neural network, the problem of difficult detection of GIS mechanical defects is solved, and efficient and accurate fault diagnosis is achieved.
Patent Information
- Application Number
- CN202510312637.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-14
- Publication Date
- 2025-08-01
AI Technical Summary
In the prior art, defects in the internal mechanical structure of gas insulated switchgear (GIS) are difficult to detect, and they are prone to failures under long-term live operation, resulting in low detection efficiency and complicated processes.
By obtaining the target one-dimensional timing vibration signal of the gas insulated switch equipment, the graph construction process is carried out to convert it into a two-dimensional map, and fault detection is used using a lightweight convolutional neural network model, and training is carried out in combination with mechanical defect types under different load currents.
It realizes efficient and accurate GIS mechanical defect diagnosis, reduces computing resource consumption, and improves real-time and adaptability of detection.
Smart Images
Figure CN120408356A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of power transmission, and particularly to a fault detection method, device, and electronic device for a Gas Insulated Switchgear (GIS). Background Art
[0002] Due to its advantages such as strong power transmission capacity, small floor area, and less maintenance work, GIS is widely used in the field of power transmission. However, it is difficult to detect internal mechanical structure defects of GIS equipment, and defects are prone to develop into faults under long-term live operation, resulting in serious consequences.
[0003] In existing research, mainly numerical feature extraction or neural network methods are used, and a one-dimensional feature vector composed of multiple features is used to solve the above-mentioned fault detection problem.
[0004] However, the existing methods have disadvantages such as a large computational amount of the network model, resulting in low detection efficiency and a complicated detection process. Summary of the Invention
[0005] Embodiments of this application provide a fault detection method, device, and electronic device for a gas insulated switchgear, so as to achieve efficient and convenient fault detection of GIS.
[0006] In a first aspect, embodiments of this application provide a fault detection method for a gas insulated switchgear, including:
[0007] Obtain a target one-dimensional time-series vibration signal of the gas insulated switchgear to be detected and the target load current where the gas insulated switchgear to be detected is located;
[0008] Perform spectral map construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectral map;
[0009] Input the target two-dimensional spectral map and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas insulated switchgear to be detected, where the fault detection model is trained based on at least one one-dimensional time-series vibration signal of the gas insulated switchgear with different mechanical defect types under different load currents for a lightweight convolutional neural network.
[0010] In one or more embodiments, the performing spectral map construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectral map includes:
[0011] Perform normalization processing on the target one-dimensional time-series vibration signal to obtain a normalized time-series vibration signal;
[0012] Perform polar coordinate transformation on the normalized time-series vibration signal to obtain time-series points;
[0013] Determine the target two-dimensional map according to the time-series points.
[0014] In one or more embodiments, the determining the target two-dimensional map according to the time-series points includes:
[0015] Determine the sum of the phase angles of the time-series points and / or the phase angle difference of the time-series points;
[0016] According to the sum of the phase angles of the time-series points and / or the phase angle difference of the time-series points, determine the Gram sum angle field corresponding to the target one-dimensional time-series vibration signal and / or the Gram difference angle field corresponding to the target one-dimensional time-series vibration signal;
[0017] Determine the Gram sum angle field or the Gram difference angle field as the target two-dimensional map.
[0018] In one or more embodiments, before inputting the target two-dimensional map and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected, the method further includes:
[0019] Obtain at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents;
[0020] Train the lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain the fault detection model.
[0021] In one or more embodiments, the training the lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain the fault detection model includes:
[0022] Perform map construction processing on at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain at least one two-dimensional map of different mechanical defect types under different load currents;
[0023] Train the lightweight convolutional neural network according to at least one two-dimensional map of different mechanical defect types under different load currents to obtain the fault detection model.
[0024] In one or more embodiments, different mechanical defect types include at least one of the following: poor contact of the disconnector, loose base of the disconnector, loose base of the bus conductor, and normal operating conditions.
[0025] Second aspect, an embodiment of the present application provides a fault detection device for a gas-insulated switchgear, including:
[0026] An acquisition module, configured to acquire a target one-dimensional time-series vibration signal of the gas-insulated switchgear to be detected and a target load current at which the gas-insulated switchgear to be detected is located;
[0027] A first processing module, configured to perform spectrogram construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrogram;
[0028] A second processing module, configured to input the target two-dimensional spectrogram and the target load current into a pre-trained fault detection model to obtain a target fault type of the gas-insulated switchgear to be detected, where the fault detection model is trained based on at least one one-dimensional time-series vibration signal of the gas-insulated switchgear with different mechanical defect types under different load currents for a lightweight convolutional neural network.
[0029] In one or more embodiments, the first processing module is specifically configured to:
[0030] Perform normalization processing on the target one-dimensional time-series vibration signal to obtain a normalized time-series vibration signal;
[0031] Perform polar coordinate conversion on the normalized time-series vibration signal to obtain time series points;
[0032] Determine the target two-dimensional spectrogram according to the time series points.
[0033] In one or more embodiments, the first processing module determines the target two-dimensional spectrogram according to the time series points, specifically configured to:
[0034] Determine the phase angle sum of the time series points and / or the phase angle difference of the time series points;
[0035] Determine the Gram sum angle field corresponding to the target one-dimensional time-series vibration signal and / or the Gram difference angle field corresponding to the target one-dimensional time-series vibration signal according to the phase angle sum of the time series points and / or the phase angle difference of the time series points;
[0036] Determine the Gram sum angle field or the Gram difference angle field as the target two-dimensional spectrogram.
[0037] In one or more embodiments, before inputting the target two-dimensional spectrogram and the target load current into the pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected, the first processing module is further configured to:
[0038] Obtain at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents;
[0039] Train the lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain the fault detection model.
[0040] In one or more embodiments, the first processing module trains the lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain the fault detection model, and specifically is used for:
[0041] Perform spectrogram construction processing on at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain at least one two-dimensional spectrogram of different mechanical defect types under different load currents;
[0042] Train the lightweight convolutional neural network according to at least one two-dimensional spectrogram of different mechanical defect types under different load currents to obtain the fault detection model.
[0043] In one or more embodiments, different mechanical defect types include at least one of the following: poor contact of disconnector, loose base of disconnector, loose base of bus conductor, and normal working condition.
[0044] In a third aspect, an embodiment of the present application provides an electronic device, including: a memory, a processor;
[0045] The memory stores computer execution instructions;
[0046] The processor executes the computer execution instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementation manners of the first aspect.
[0047] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer execution instructions are stored, and when the computer execution instructions are executed by a processor, they are used to implement the above first aspect and / or various possible implementation manners of the first aspect.
[0048] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the above first aspect and / or various possible implementation manners of the first aspect.
[0049] An embodiment of the present application provides a fault detection method, device, and electronic device for a gas-insulated switchgear. The method includes obtaining a target one-dimensional time-series vibration signal of the gas-insulated switchgear to be detected and the target load current at which the gas-insulated switchgear to be detected is located; performing spectrogram construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrogram; inputting the target two-dimensional spectrogram and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected. The fault detection model is trained by using at least one one-dimensional time-series vibration signal of the gas-insulated switchgear with different mechanical defect types under different load currents for a lightweight convolutional neural network. In this technical solution, the one-dimensional defect vibration signal is converted into a two-dimensional spectrogram, and the two-dimensional spectrogram contains rich defect features, which can amplify the features of the defect signal and can be used as the input for defect diagnosis. The lightweight convolutional neural network model has a good fault diagnosis effect and can accurately and reliably diagnose GIS mechanical defects. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] The accompanying drawings are incorporated herein and form a part of this specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.
[0051] Figure 1 Schematic flow chart of fault detection of gas-insulated switchgear provided by an embodiment of the present application Figure 1 ;
[0052] Figure 2 Schematic diagram of the conversion of one-dimensional time-series vibration signal provided by an embodiment of the present application;
[0053] Figure 3 Schematic flow chart of fault detection of gas-insulated switchgear provided by an embodiment of the present application Figure 2 ;
[0054] Figure 4 Schematic flow chart of fault detection of gas-insulated switchgear provided by an embodiment of the present application Figure 3 ;
[0055] Figure 5 Schematic diagram of the two-dimensional spectrogram corresponding to the mechanical defect type provided by an embodiment of the present application;
[0056] Figure 6 Schematic flow chart of fault detection of gas-insulated switchgear provided by an embodiment of the present application Figure 4 ;
[0057] Figure 7 Schematic diagram of fault detection of gas-insulated switchgear provided by an embodiment of the present application;
[0058] Figure 8Schematic structural diagram of the fault detection device for the gas-insulated switchgear provided in the embodiment of the present application;
[0059] Figure 9 Schematic structural diagram of the electronic device provided in the embodiment of the present application.
[0060] Through the above-mentioned drawings, specific embodiments of the present application have been shown, and more detailed descriptions will be provided hereinafter. These drawings and textual descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. Detailed implementation manners
[0061] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all the implementation manners consistent with the present application. On the contrary, they are merely examples of the devices and methods consistent with some aspects of the present application as detailed in the appended claims.
[0062] Due to its advantages such as strong power transmission capacity, small floor area, and less maintenance work, GIS is widely used in the field of power transmission. However, it is difficult to detect the internal mechanical structure defects of GIS equipment, and the defects are likely to develop into faults under the long-term live operation state, resulting in serious consequences. Since the mechanical structure defects of GIS have characteristics such as weak features, strong latency, and unobvious features, the research on the mechanical defect diagnosis of GIS equipment has attracted much attention.
[0063] In the prior art, machine learning can automatically adjust the structural parameters, change its mapping relationship, and has strong generalization ability, and has been widely used in the field of fault diagnosis. For example, the feature vectors of GIS acoustic wave data are extracted, and a fault diagnosis algorithm based on an online sequential extreme learning machine is adopted to achieve accurate diagnosis of mechanical defects.
[0064] However, in the above implementation, mainly numerical feature extraction or neural network methods are used to solve the above-mentioned fault detection problem by using a one-dimensional feature vector composed of multiple features, which has disadvantages such as a large computational amount of the network model, resulting in low detection efficiency and complicated detection process.
[0065] Based on the above existing technical problems, the technical concept of the inventor is as follows: The mechanical defects of GIS are the main causes of equipment failure. Accurately and efficiently identifying the defect types is crucial for ensuring the safe operation of the equipment. Aiming at the technical problem that it is difficult to select characteristic parameters due to the high complexity of GIS mechanical defect signals, the inventor found in the actual implementation that the Gramian angular field method shows good applicability in the image transformation of one-dimensional vibration signals, can amplify the characteristics of defect signals, and can be used as the input for defect diagnosis. The SqueezeNet lightweight convolutional neural network model has a good fault diagnosis effect, providing a new way for accurately and reliably diagnosing GIS mechanical defects.
[0066] The following uses specific embodiments to detail the technical solutions of the present application and how the technical solutions of the present application solve the above technical problems. These several specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below with reference to the accompanying drawings.
[0067] Figure 1 Flow schematic of fault detection for gas-insulated switchgear provided in the embodiments of the present application Figure 1 , as Figure 1 shown, the method includes:
[0068] Step 11: Obtain the target one-dimensional time-series vibration signal of the gas-insulated switchgear to be detected and the target load current where the gas-insulated switchgear to be detected is located.
[0069] In this step, in order to detect the fault type of the gas-insulated switchgear to be detected, it is necessary to first collect the vibration signal within the preset duration of the gas-insulated switchgear to be detected, denoted as the target one-dimensional time-series vibration signal.
[0070] Since the time-series vibration signal of the gas-insulated switchgear to be detected is different under different load currents, therefore, the load current where the gas-insulated switchgear to be detected is located is obtained simultaneously, denoted as the target load current.
[0071] Optionally, the target one-dimensional time-series vibration signal can be obtained based on acceleration devices such as vibration sensors. This sensor can detect the vibration signal generated during the operation of the gas-insulated switchgear and convert it into an electrical signal, which is transmitted to the execution entity of the present application. Of course, the execution entity of the present application can also include this sensor.
[0072] Step 12: Perform spectrum construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrum.
[0073] In this step, the Gramian Angular Field (GAF) can be used to construct a spectrogram for the target one-dimensional time-series vibration signal under the target load current, obtaining a two-dimensional vibration spectrogram, denoted as the target two-dimensional spectrogram.
[0074] Among them, the Gramian Angular Field is a method of converting a one-dimensional sequence into a two-dimensional image, which can not only completely retain the signal information but also maintain the signal's dependence on time. Specifically, the time series in the Cartesian coordinate system can be encoded in polar coordinates, and then the Gram matrix can be generated through trigonometric operations, thereby converting the one-dimensional time series into a two-dimensional image.
[0075] Optionally, Figure 2 is a conversion schematic diagram of the one-dimensional time-series vibration signal provided by the embodiment of the present application. As Figure 2 shown, it is a conversion schematic diagram of the Gramian angular difference field (GADF), and it can also be implemented based on the Gramian angular summation field (GASF).
[0076] The target one-dimensional time-series vibration signal is Figure 2 on the left side, including: the vibration acceleration collected in the time period of 0 - 500, which is converted to Figure 2 on the right side.
[0077] Step 13: Input the target two-dimensional spectrogram and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected;
[0078] Among them, the fault detection model is trained by using at least one one-dimensional time-series vibration signal of the gas-insulated switchgear under different load currents and different mechanical defect types on a lightweight convolutional neural network.
[0079] In this step, after obtaining the target two-dimensional spectrogram and the target load current as above, input the target two-dimensional spectrogram and the target load current into the fault detection model. Then, the fault detection model processes the target two-dimensional spectrogram and the target load current to obtain the fault type of the gas-insulated switchgear to be detected, denoted as the target fault type.
[0080] An embodiment of the present application provides a fault detection method for a gas-insulated switchgear. By obtaining the target one-dimensional time-series vibration signal of the gas-insulated switchgear to be detected and the target load current at which the gas-insulated switchgear to be detected is located; performing a spectrogram construction process on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrogram; inputting the target two-dimensional spectrogram and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected. The fault detection model is trained by using at least one one-dimensional time-series vibration signal of the gas-insulated switchgear with different mechanical defect types under different load currents for a lightweight convolutional neural network. In this technical solution, the one-dimensional defect vibration signal is converted into a two-dimensional spectrogram, and the two-dimensional spectrogram contains rich defect features, which can amplify the features of the defect signal and can be used as the input for defect diagnosis. The lightweight convolutional neural network model has a good fault diagnosis effect and can accurately and reliably diagnose GIS mechanical defects.
[0081] Based on the above embodiments, Figure 3 is a schematic flow chart of the fault detection of the gas-insulated switchgear provided by the embodiment of the present application Figure 2 , as Figure 3 shown, step 12 may include:
[0082] Step 31: Perform normalization processing on the target one-dimensional time-series vibration signal to obtain a normalized time-series vibration signal;
[0083] In this step, after obtaining the target one-dimensional time-series vibration signal, perform normalization processing on the target one-dimensional time-series vibration signal.
[0084] Optionally, taking the target one-dimensional time-series vibration signal X = {x1, x2,..., xn} as an example, where n is the number corresponding to the time series.
[0085] Perform normalization processing based on the following formula to obtain a normalized time-series vibration signal.
[0086]
[0087] where i is any one of 1 - n, is the normalized vibration signal corresponding to the vibration signal x i corresponding to.
[0088] Step 32: Perform polar coordinate transformation on the normalized time-series vibration signal to obtain time series points;
[0089] In this step, after obtaining the time-series vibration signal through the above formula processing, perform polar coordinate transformation processing on the time-series vibration signal to obtain time series points for constructing the target two-dimensional spectrogram.
[0090] For example, the normalized X is denoted as By transforming the normalized time - series vibration signal into the polar coordinate system, polar coordinate conversion is achieved, and time - series points are obtained.
[0091] Optionally, polar coordinate conversion processing is performed based on the following formula to obtain time - series points.
[0092]
[0093] where φi is the polar angle after the inverse cosine function transformation; ri is the polar radius.
[0094] Step 33: Determine the target two - dimensional spectrum according to the time - series points.
[0095] In this step, after obtaining the time series above, the sum and / or difference of the phase angles can be calculated based on the time series, so as to construct the target two - dimensional spectrum.
[0096] Optionally, a possible implementation of step 33 can be:
[0097] Step 1: Determine the sum of the phase angles of the time - series points and / or the difference of the phase angles of the time - series points;
[0098] The sum of the phase angles, such as cos(φ1 + φ1); that is, the sum of the phase angles of φ1 itself; cos(φ1 + φ2); that is, the sum of the phase angles of φ1 and φ2;
[0099] The difference of the phase angles, such as sin(φ1 + φ1); that is, the sum of the phase angles of φ1 itself; sin(φ1 + φ2); that is, the sum of the phase angles of φ1 and φ2.
[0100] Step 2: Determine the Gram sum - angle field corresponding to the target one - dimensional time - series vibration signal and / or the Gram difference - angle field corresponding to the target one - dimensional time - series vibration signal according to the sum of the phase angles of the time - series points and / or the difference of the phase angles of the time - series points;
[0101] Since there are two calculation methods of the sum and / or difference of the phase angles, in actual calculation, only one of them needs to be executed.
[0102] That is, the Gram angle field can be divided into two forms: GASF and GADF.
[0103] For GASF:
[0104]
[0105] For GADF:
[0106]
[0107] In the present application, the above Figure 2Using the calculation method of Gram difference angle field, in the above formula, φn is the polar angle after the inverse cosine function transformation.
[0108] Step 3: Determine the Gram sum angle field or Gram difference angle field as the target two-dimensional map.
[0109] Use the determined Gram sum angle field or Gram difference angle field as the target two-dimensional map to provide to step 13.
[0110] The fault detection method for the gas-insulated switchgear provided by the embodiment of the present application normalizes the target one-dimensional time-series vibration signal to obtain the normalized time-series vibration signal, performs polar coordinate transformation on the normalized time-series vibration signal to obtain time-series points, and determines the target two-dimensional map according to the time-series points. In this technical solution, by normalizing the target one-dimensional time-series vibration signal, the influence of different signal amplitudes and dimensions can be eliminated, enabling comparison of the various features of the signal under the same standard, thereby improving the accuracy of subsequent analysis. Then, performing polar coordinate transformation on the normalized time-series vibration signal can convert the one-dimensional signal into a two-dimensional data form, excavate richer spatial features in the signal, enhance the accuracy of fault detection, and finally generate the target two-dimensional map according to the converted time-series points for subsequent fault detection processes.
[0111] Based on the above embodiments, Figure 4 is the flow diagram of the fault detection of the gas-insulated switchgear provided by the embodiment of the present application Figure 3 As Figure 4 shown, before step 13, it may further include:
[0112] Step 41: Obtain at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents;
[0113] In this step, in order to obtain a more accurate and comprehensive fault detection model, one-dimensional time-series vibration signals of various mechanical defect types of the gas-insulated switchgear under different load currents can be obtained as the training set and test set of the fault detection model.
[0114] In a possible implementation, load currents corresponding to 600A, 1200A, 1800A, and 2400A are used, and for each current, different mechanical defect types can be constructed, such as poor contact of the disconnector, loose base of the disconnector, loose base of the bus conductor, and normal operating conditions. And collect the one-dimensional time-series vibration signals under the above conditions.
[0115] Step 42: Train a lightweight convolutional neural network based on at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain a fault detection model.
[0116] In this step, after obtaining at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents, train a lightweight convolutional neural network based on at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents until the test results meet the preset requirements to obtain a fault detection model for actual detection.
[0117] Optionally, a possible implementation of Step 42 is as follows:
[0118] Step 1: Perform spectrogram construction processing on at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain at least one two-dimensional spectrogram of different mechanical defect types under different load currents;
[0119] In this implementation, similar to the implementation principle disclosed in the above embodiment, for each load current and each mechanical defect type corresponding to the load current, perform spectrogram construction processing on the one-dimensional time-series vibration signal of the mechanical defect type to obtain a two-dimensional spectrogram corresponding to the one-dimensional time-series vibration signal of the mechanical defect type.
[0120] Exemplarily, different mechanical defect types include at least one of the following: poor contact of disconnector, loose base of disconnector, loose base of bus conductor, and normal working condition.
[0121] Figure 5 Schematic diagram of the two-dimensional spectrogram corresponding to the mechanical defect type provided in the embodiment of the present application, as Figure 5 shown, taking the GIS device under 600A current in four operating conditions (mechanical defect types) as an example, a below is the normal working condition, b is, c is poor contact of disconnector and loose base of disconnector, d is loose base of bus conductor.
[0122] Step 2: Train a lightweight convolutional neural network based on at least one two-dimensional spectrogram of different mechanical defect types under different load currents to obtain a fault detection model.
[0123] In this implementation, for each load current, input at least one two-dimensional spectrogram (such as 336) under the load current into the SqueezeNet model respectively, and divide the training set (252) and the test set (84) according to a preset ratio, such as a ratio of 3:1. The overall diagnostic accuracy of the test set samples obtained after model training is shown in Table 1.
[0124] Among the 84 sample spectrograms in the test set, at most 4 spectrograms are diagnosed incorrectly. The diagnostic accuracies at four load currents of 600A, 1200A, 1800A, and 2400A are 95.24%, 98.81%, 98.81%, and 96.43% respectively, and the classification effect is relatively good.
[0125] That is, Table 1 shows the accuracies of the GIS mechanical defect diagnosis models under different load currents provided by the embodiments of the present application.
[0126] Table 1
[0127]
[0128]
[0129] The fault detection method for gas-insulated switchgear provided by the embodiments of the present application obtains at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents, and trains a lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain a fault detection model. In this technical solution, obtaining at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents can comprehensively reflect the operating state of the equipment under different working conditions and defect types. These signals, as input data, help to extract time-series information related to fault characteristics, thereby providing rich samples for model training. Then, by applying these time-series signals to train a lightweight convolutional neural network, the spatio-temporal characteristics in the signals can be effectively learned, and the fault recognition ability of the model can be optimized. The use of a lightweight convolutional neural network can reduce the consumption of computing resources while maintaining a high recognition accuracy, and improve the real-time performance and adaptability of the model.
[0130] Figure 6 is a schematic flow chart of the fault detection of the gas-insulated switchgear provided by the embodiments of the present application Figure 4 This method includes:
[0131] Among them, in combination with Figure 7 for illustration, Figure 7 is a schematic diagram of the fault detection of the gas-insulated switchgear provided by the embodiments of the present application.
[0132] [[ID=31]]Step 61: Convert the one-dimensional time-series vibration signal of the GIS device into a two-dimensional spectrogram;
[0133] Step 62: Construct a GIS device mechanical defect diagnosis model (fault detection model) based on a lightweight convolutional neural network;
[0134] Step 63: Calculate the diagnosis result of the GIS device mechanical defect.
[0135] Among them, the two-dimensional spectrum generated in step 61 is used as the input of the SqueezeNet lightweight convolutional neural network diagnostic model. After model training and verification by the test set, the mechanical defect diagnosis result is obtained.
[0136] Based on Figure 7 , in the training of the fault detection model, it can be trained based on the lightweight convolution (i.e., SqueezeNet) neural network. Specifically, since it can keep the model performance similar to AlexNet while only having 1 / 60 of the parameters of AlexNet, SqueezeNet has the following three characteristics:
[0137] (1) Since smaller convolutional kernels have fewer computational parameters, SqueezeNet pioneered replacing some 3×3 convolutional kernels in the convolutional neural network with 1×1 convolutional kernels.
[0138] (2) Replace the convolutional layer with a fire module structure composed of a squeeze layer and an expansion layer: the input features enter the squeeze layer composed of 1×1 convolutional kernels to reduce the number of feature channels, then enter the expansion layer composed of 1×1 convolutional kernels and 3×3 convolutional kernels, and finally splice the output features of the 1×1 convolutional kernels and 3×3 convolutional kernels together, reducing the number of parameters of the entire network.
[0139] (3) Delay the pooling operation and retain a larger feature map, so it has more feature information. This neural network consists of 68 layers, and the total number of learnable parameters is 727.6k. It mainly includes an image input layer, a two-dimensional convolutional layer, an activation function rectified linear unit (ReLU), a depth concatenation layer, a two-dimensional max pooling layer, a two-dimensional average pooling layer, a classification output layer, etc.
[0140] The fault detection method for gas-insulated switchgear provided by the embodiments of the present application has a similar implementation principle and technical effect to the above embodiments, which will not be elaborated here.
[0141] The following is the description of the device embodiments provided by the embodiments of the present application.
[0142] Figure 8 It is a schematic structural diagram of the fault detection device for gas-insulated switchgear provided by the embodiments of the present application. As Figure 8 shown, the fault detection device for gas-insulated switchgear provided by this embodiment includes:
[0143] An acquisition module 81, configured to acquire the target one-dimensional time-series vibration signal of the gas-insulated switchgear to be detected and the target load current where the gas-insulated switchgear to be detected is located;
[0144] A first processing module 82, configured to perform spectrum construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrum;
[0145] The second processing module 83 is configured to input the target two-dimensional map and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected. The fault detection model is trained by using at least one one-dimensional time-series vibration signal of the gas-insulated switchgear with different mechanical defect types under different load currents to train a lightweight convolutional neural network.
[0146] In one or more embodiments, the first processing module 81 is specifically configured to:
[0147] Perform normalization processing on the target one-dimensional time-series vibration signal to obtain a normalized time-series vibration signal;
[0148] Perform polar coordinate conversion on the normalized time-series vibration signal to obtain time series points;
[0149] Determine the target two-dimensional map according to the time series points.
[0150] In one or more embodiments, the first processing module 81, according to the time series points, determines the target two-dimensional map, specifically configured to:
[0151] Determine the phase angle sum of the time series points and / or the phase angle difference of the time series points;
[0152] According to the phase angle sum of the time series points and / or the phase angle difference of the time series points, determine the Gram sum angle field corresponding to the target one-dimensional time-series vibration signal and / or the Gram difference angle field corresponding to the target one-dimensional time-series vibration signal;
[0153] Determine the Gram sum angle field or the Gram difference angle field as the target two-dimensional map.
[0154] In one or more embodiments, before inputting the target two-dimensional map and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected, the first processing module 81 is further configured to:
[0155] Obtain at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents;
[0156] Train a lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain a fault detection model.
[0157] In one or more embodiments, the first processing module 81, according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents, trains a lightweight convolutional neural network to obtain a fault detection model, specifically configured to:
[0158] Construct a spectrogram for at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain at least one two-dimensional spectrogram of different mechanical defect types under different load currents;
[0159] Train a lightweight convolutional neural network based on at least one two-dimensional spectrogram of different mechanical defect types under different load currents to obtain a fault detection model.
[0160] In one or more embodiments, different mechanical defect types include at least one of the following: poor contact of the disconnector, loose base of the disconnector, loose base of the busbar conductor, and normal operating conditions.
[0161] The fault detection device of the gas-insulated switchgear provided in this embodiment can execute the fault detection method of the gas-insulated switchgear provided in the above method embodiment. The implementation principle and technical effect are similar, and will not be elaborated here in this embodiment.
[0162] Figure 9 This is a schematic structural diagram of an electronic device provided in an embodiment of the present application. As Figure 9 shown, the electronic device provided in this embodiment includes:
[0163] At least one processor 91 and a memory 92.
[0164] Optionally, the device further includes a communication component 93. Among them, the processor 91, the memory 92, and the communication component 93 are connected through a bus 94.
[0165] In a specific implementation process, at least one processor 91 executes the computer execution instructions stored in the memory 92, so that at least one processor 91 executes the above method.
[0166] The specific implementation process of the processor 91 can be referred to in the above method embodiment. The implementation principle and technical effect are similar, and will not be elaborated here in this embodiment.
[0167] In the above embodiment, it should be understood that the processor may be a central processing unit (English: Central Processing Unit, abbreviated: CPU), or other general-purpose processors, digital signal processors (English: Digital Signal Processor, abbreviated: DSP), application specific integrated circuits (English: Application Specific Integrated Circuit, abbreviated: ASIC), etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the invention can be directly implemented by the hardware processor to complete, or implemented by a combination of hardware and software modules in the processor.
[0168] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-volatile Memory, NVM), such as at least one disk memory.
[0169] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the buses in the drawings of this application are not limited to only one bus or one type of bus.
[0170] This application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.
[0171] This application also provides a computer-readable storage medium, in which computer-executable instructions are stored, and when the processor executes the computer-executable instructions, the above method is implemented.
[0172] The above-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk or an optical disk. The readable storage medium can be any available medium accessible by a general-purpose or special-purpose computer.
[0173] An exemplary readable storage medium is coupled to the processor, enabling the processor to read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an Application Specific Integrated Circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in a device.
[0174] The division of units is merely a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the couplings or direct couplings or communication connections shown or discussed among each other can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.
[0175] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0176] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.
[0177] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks or optical discs and other various media that can store program codes.
[0178] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When this program is executed, it executes the steps including the above method embodiments; and the aforementioned storage medium includes: ROMs, RAMs, magnetic disks or optical discs and other various media that can store program codes.
[0179] Finally, it should be noted that: After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily conceive of other embodiments of the present invention. The present invention is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include common general knowledge or conventional technical means in the technical field not disclosed in the present invention. It is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.
Claims
1. A fault detection method for a gas-insulated switchgear, characterized in that, Including: Obtain a target one-dimensional time-series vibration signal of the gas-insulated switchgear to be detected and the target load current at which the gas-insulated switchgear to be detected is located; Perform spectrum construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrum; Input the target two-dimensional spectrum and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected, where the fault detection model is trained by using at least one one-dimensional time-series vibration signal of the gas-insulated switchgear with different mechanical defect types under different load currents for a lightweight convolutional neural network.
2. The method according to claim 1, wherein The performing spectrum construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrum includes: Perform normalization processing on the target one-dimensional time-series vibration signal to obtain a normalized time-series vibration signal; Perform polar coordinate transformation on the normalized time-series vibration signal to obtain time series points; Determine the target two-dimensional spectrum according to the time series points.
3. The method according to claim 2, characterized in that The determining the target two-dimensional spectrum according to the time series points includes: Determine the sum of phase angles of the time series points and / or the phase angle difference of the time series points; Determine the Gram sum angle field corresponding to the target one-dimensional time-series vibration signal and / or the Gram difference angle field corresponding to the target one-dimensional time-series vibration signal according to the sum of phase angles of the time series points and / or the phase angle difference of the time series points; Determine the Gram sum angle field or the Gram difference angle field as the target two-dimensional spectrum.
4. The method according to any one of claims 1 to 3, characterized in that, Before inputting the target two-dimensional spectrum and the target load current into a pre-trained fault detection model to obtain the target fault type of the gas-insulated switchgear to be detected, the method further includes: Obtain at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents; Train the lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain the fault detection model.
5. The method according to claim 4, wherein The training the lightweight convolutional neural network according to at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain the fault detection model includes: Perform spectrum construction processing on at least one one-dimensional time-series vibration signal of different mechanical defect types under different load currents to obtain at least one two-dimensional spectrum of different mechanical defect types under different load currents; Train the lightweight convolutional neural network according to at least one two-dimensional spectrum of different mechanical defect types under different load currents to obtain the fault detection model.
6. The method according to any one of claims 1 to 3, characterized in that, Different mechanical defect types include at least one of the following: poor contact of the disconnector, loose base of the disconnector, loose base of the bus conductor, and normal operating conditions.
7. A fault detection device for a gas-insulated switchgear, characterized in that, Including: An acquisition module, configured to obtain a target one-dimensional time-series vibration signal of the gas-insulated switchgear to be detected and the target load current at which the gas-insulated switchgear to be detected is located; A first processing module, configured to perform spectrum construction processing on the target one-dimensional time-series vibration signal to obtain a target two-dimensional spectrum; A second processing module, configured to input the target two-dimensional spectrum and the target load current into a pre-trained fault detection model to obtain a target fault type of the gas-insulated switchgear to be detected, where the fault detection model is trained based on at least one one-dimensional time-series vibration signal of different mechanical defect types of the gas-insulated switchgear under different load currents for a lightweight convolutional neural network.
8. An electronic device, characterized in that, Comprising: A memory and a processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor executes the method according to any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, Computer-executable instructions are stored in the computer-readable storage medium, and when the computer-executable instructions are executed by a processor, they are used to implement the method according to any one of claims 1-6.
10. A computer program product, characterized in that, Including a computer program, which is used to implement the method according to any one of claims 1-6 when executed by a processor.