Isolation switch mechanical fault detection method, device, equipment, medium and product

Abstract

The embodiment of the invention provides an isolation switch mechanical fault detection method, device and equipment, a medium and a product, and is applied to the technical field of power grid system fault diagnosis. The method comprises the following steps: acquiring time sequence data of multiple dimensions generated by the disconnecting switch within a preset time period; wherein the time sequence data of the multiple dimensions comprises mechanical vibration data, current data and environment temperature data of the disconnecting switch; feature extraction is carried out on the time sequence data of each dimension, and features of the time sequence data of multiple dimensions are fused to obtain fused time sequence features; inputting the fusion time sequence characteristics into a multi-dimensional fault detection model to obtain a predicted fault detection result corresponding to the disconnecting switch; wherein the multi-dimensional fault detection model is a multi-dimensional fault detection model obtained by training a time sequence prediction model based on feature data of the disconnecting switch in multiple dimensions. The method is used for achieving the effect of improving the fault detection efficiency of the disconnecting switch.

Description

Technical Field

[0001] This application relates to the technical field of power grid system fault diagnosis, and particularly to a detection method, device, equipment, medium and product for mechanical faults of disconnectors. Background Art

[0002] In a substation, there are a large number of disconnectors, which are used to achieve power isolation, switching operations and form visible disconnection points inside the substation; the mechanical performance of the disconnectors can directly affect the safety and reliability of the substation power system. In order to ensure the normal operation of the substation, it is necessary to conduct regular fault detection and maintenance on the disconnectors.

[0003] In the prior art, the main method for fault detection of disconnectors is: creating a feature comparison library based on historical fault diagnosis data, and comparing the collected operation data with the feature data in the feature comparison library to determine whether there is a fault in the current disconnector and the type of the existing fault.

[0004] Since the prior art mainly uses known fault cases for feature comparison, some faults are missed, resulting in the technical problem of low efficiency in detecting disconnector faults in the prior art. Summary of the Invention

[0005] The embodiments of this application provide a detection method, device, equipment, medium and product for mechanical faults of disconnectors, so as to achieve the effect of improving the efficiency of detecting disconnector faults.

[0006] In a first aspect, the embodiments of this application provide a detection method for mechanical faults of disconnectors, including:

[0007] Collecting multi-dimensional time series data generated by the disconnector within a preset time period; wherein, the multi-dimensional time series data includes: mechanical vibration data, current data and ambient temperature data of the disconnector;

[0008] Extracting features for each dimension of the time series data, and fusing the features of the multi-dimensional time series data to obtain fused time series features;

[0009] Inputting the fused time series features into a multi-dimensional fault detection model to obtain a predicted fault detection result corresponding to the disconnector;

[0010] Wherein, the multi-dimensional fault detection model is a multi-dimensional fault detection model obtained by training a time series prediction model based on the feature data of the disconnector in multiple dimensions.

[0011] In a possible implementation manner, before collecting the multi-dimensional time series data generated by the disconnector within a preset time period, the method further includes:

[0012] Perform data processing based on historical fault detection data to obtain a sample training data set and a sample validation data set;

[0013] Perform model training on a preset initial model based on the sample training data set to obtain a model to be determined;

[0014] Perform anomaly verification on the model to be determined based on the sample validation data set. When the anomaly verification result of the model to be determined passes the verification, determine the model to be determined as a multi-dimensional fault detection model.

[0015] In a possible implementation manner, performing outlier verification on the model to be determined based on the sample validation set. When the outlier verification result of the model to be determined passes the verification, determining the model to be determined as a multi-dimensional fault detection model includes:

[0016] Calculate the anomaly score of the model to be determined based on the sample validation data set; wherein, the anomaly score is used to indicate the probability of the occurrence of an abnormal prediction result in the model to be determined;

[0017] When the anomaly score belongs to a preset anomaly threshold range, determine the model to be determined as a multi-dimensional fault detection model;

[0018] When the anomaly score does not belong to the preset anomaly threshold range, update the model to be determined based on the abnormal data and its corresponding abnormal prediction results in the sample validation data set until the anomaly score of the model to be determined belongs to the preset anomaly threshold range.

[0019] In a possible implementation manner, collect multi-dimensional time series data generated by the disconnector within a preset time period, including:

[0020] Connect multi-source sensors pre-deployed on the disconnector;

[0021] For different types of time series data, record the numerical changes of the multi-source sensors within the preset time period to obtain multi-dimensional time series data.

[0022] In a possible implementation manner, perform feature extraction on the time series data of each dimension and fuse the features corresponding to the time series data of multiple dimensions to obtain fused time series features, including:

[0023] Perform data processing on the time series data of each dimension to obtain time series data of multiple dimensions with time alignment;

[0024] Perform feature extraction on the time series data with time alignment to obtain time series data features corresponding to the time series data of each dimension;

[0025] Fuse the time series features corresponding to the time series data of multiple dimensions to obtain fused time series features.

[0026] In a possible implementation, inputting the fused temporal features into a multi-dimensional fault detection model to obtain the predicted fault detection result corresponding to the disconnector, including:

[0027] Performing weighted processing on the temporal features of different dimensions in the fused temporal features based on the attention mechanism to obtain a joint feature matrix;

[0028] Inputting the joint feature matrix into the multi-dimensional fault detection model to obtain the predicted fault detection result;

[0029] Among them, the predicted fault detection result includes the predicted fault type, predicted fault location, predicted fault occurrence probability, and predicted fault occurrence time.

[0030] In a possible implementation, after inputting the fused temporal features into the multi-dimensional fault detection model to obtain the predicted fault detection result corresponding to the disconnector, the method further includes:

[0031] Matching the predicted fault type and predicted fault location in the predicted fault detection result with a preset knowledge base, and when the matching is successful, extracting the disposal strategy corresponding to the predicted fault detection result from the preset knowledge base;

[0032] Determining the priority corresponding to the predicted fault detection result based on the predicted fault occurrence probability and predicted fault occurrence time in the predicted fault detection result;

[0033] Generating a fault solution corresponding to the predicted fault detection result based on the disposal strategy and the priority;

[0034] Among them, the preset knowledge base contains the processing strategies corresponding to different predicted fault types and different predicted fault locations of the disconnector.

[0035] In a second aspect, an embodiment of the present application provides a detection device for mechanical faults of a disconnector, including:

[0036] An acquisition module, configured to collect multi-dimensional temporal data generated by the disconnector within a preset time period; among them, the multi-dimensional temporal data includes: mechanical vibration data, current data, and ambient temperature data of the disconnector;

[0037] A first processing module, configured to extract features for each dimension of the temporal data and fuse the features of the multi-dimensional temporal data to obtain fused temporal features;

[0038] A second processing module, configured to input the fused temporal features into a multi-dimensional fault detection model to obtain the predicted fault detection result corresponding to the disconnector;

[0039] Among them, the multi-dimensional fault detection model is a multi-dimensional fault detection model obtained by training a time series prediction model based on the feature data of the disconnector in multiple dimensions.

[0040] In a possible implementation manner, the device further includes a third processing module, configured to:

[0041] Perform data processing on the historical fault detection data to obtain a sample training data set and a sample verification data set;

[0042] Perform model training on a preset initial model based on the sample training data set to obtain a model to be determined;

[0043] Perform anomaly verification on the model to be determined based on the sample verification data set. When the anomaly verification result of the model to be determined passes the verification, determine the model to be determined as the multi-dimensional fault detection model.

[0044] In a possible implementation manner, the third processing module is further configured to:

[0045] Calculate the anomaly score of the model to be determined based on the sample verification data set; wherein, the anomaly score is used to indicate the probability of the occurrence of an abnormal prediction result in the model to be determined;

[0046] When the anomaly score belongs to a preset anomaly threshold range, determine the model to be determined as the multi-dimensional fault detection model;

[0047] When the anomaly score does not belong to the preset anomaly threshold range, update the model to be determined based on the abnormal data and its corresponding abnormal prediction result in the sample verification data set until the anomaly score of the model to be determined belongs to the preset anomaly threshold range.

[0048] In a possible implementation manner, the acquisition module is further configured to:

[0049] Connect to multi-source sensors pre-deployed on the disconnector;

[0050] For different types of time series data, record the numerical changes of the multi-source sensors within a preset time period to obtain time series data in multiple dimensions.

[0051] In a possible implementation manner, the first processing module is further configured to:

[0052] Perform data processing on the time series data in each dimension to obtain time series data in multiple dimensions with time series alignment;

[0053] Perform feature extraction on the time series data with time series alignment to obtain time series data features corresponding to the time series data in each dimension;

[0054] Fuse the time series features corresponding to the time series data in multiple dimensions to obtain fused time series features.

[0055] In a possible implementation, the first processing module is further configured to:

[0056] Perform weighted processing on the time series features of different dimensions in the fused time series features based on the attention mechanism to obtain a joint feature matrix;

[0057] Input the joint feature matrix into a multi-dimensional fault detection model to obtain a predicted fault detection result;

[0058] Wherein, the predicted fault detection result includes a predicted fault type, a predicted fault location, a predicted fault occurrence probability, and a predicted fault occurrence time.

[0059] In a possible implementation, the second processing module is further configured to:

[0060] Match the predicted fault type and the predicted fault location in the predicted fault detection result with a preset knowledge base, and when the match is successful, extract the disposal strategy corresponding to the predicted fault detection result from the preset knowledge base;

[0061] Determine the priority corresponding to the predicted fault detection result based on the predicted fault occurrence probability and the predicted fault occurrence time in the predicted fault detection result;

[0062] Generate a fault solution corresponding to the predicted fault detection result based on the disposal strategy and the priority;

[0063] Wherein, the preset knowledge base includes processing strategies corresponding to different predicted fault types and different predicted fault locations of the disconnector.

[0064] In a third aspect, an embodiment of the present application provides an electronic device, including: a memory, a processor;

[0065] The memory stores computer-executable instructions;

[0066] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0067] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored, and when the computer-executable instructions are executed by a processor, they are used to implement the above first aspect and / or various possible implementations of the first aspect.

[0068] 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 implementations of the first aspect.

[0069] The detection method, device, equipment, medium and product for mechanical faults of disconnectors provided by the embodiments of the present application collect time-series data of multiple dimensions generated by the disconnector within a preset time period, and obtain mechanical vibration data, current data and ambient temperature data generated during the operation of the disconnector; extract features for the time-series data of each dimension, fuse the features of the time-series data of multiple dimensions to obtain fused time-series features; the fused time-series features contain the time-series data features of the disconnector in multiple dimensions; input the fused time-series features into a pre-trained multi-dimensional fault detection model to obtain the predicted fault detection result of the multi-dimensional fault detection model for the fused time-series features; thus, it is determined whether the disconnector has a mechanical fault and the type of the existing mechanical fault according to the preset fault detection result. Compared with the method of comparing with known fault cases in the prior art, the present application uses a multi-dimensional fault detection model to predict faults of the disconnector, which can not only analyze the existing faults of the disconnector, but also predict the possible faults of the disconnector, so as to improve the efficiency of disconnector fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present application and used together with the specification to explain the principles of the present application.

[0071] Figure 1 Schematic flow chart of the detection method for mechanical faults of the disconnector provided by the present application Figure 1 ;

[0072] Figure 2 Schematic flow chart of the training method for the multi-dimensional fault detection model provided by the present application;

[0073] Figure 3 Schematic flow chart of the detection method for mechanical faults of the disconnector provided by the present application Figure 2 ;

[0074] Figure 4 Schematic structural diagram of the detection device for mechanical faults of the disconnector provided by the present application;

[0075] Figure 5 Schematic structural diagram of the electronic device provided by the present application.

[0076] Through the above-mentioned accompanying drawings, specific embodiments of the present application have been shown, and there will be more detailed descriptions hereinafter. These drawings and text descriptions are not intended to limit the scope of the concept of the present application in any way, but to explain the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0077] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0078] In the prior art, the main methods for fault detection of disconnectors in a substation are as follows: Regularly check the operation status of the disconnector through manual inspection to detect faults in the disconnector, thereby determining the corresponding fault type, and regularly giving a maintenance plan; or use sensors to collect the operation data of the disconnector, extract the features in the operation data and compare them with the features in the historical feature library to determine the fault type corresponding to the disconnector, and give the solution corresponding to the current fault type according to the historical fault solution, and push the solution to the maintenance organization or personnel to achieve fault detection and maintenance of the disconnector.

[0079] However, in the prior art, the time period for fault detection of disconnectors by manual inspection is long, and real-time fault detection cannot be achieved. At the same time, the method of feature comparison using the historical feature library requires a large amount of feature data to be stored in the feature library. If the corresponding fault type of the current disconnector is not matched in the feature library, manual inspection is also required to determine the fault type. Therefore, there is a technical problem of low efficiency in fault detection of disconnectors in the prior art.

[0080] In view of the above technical problems, the present application proposes the following technical concept: Use machine learning to detect faults. Specifically: Collect the time-series data of multiple dimensions of the disconnector within a preset time period, aiming to obtain the real-time operation data of the disconnector; Extract and fuse the features corresponding to the time-series data of multiple dimensions to obtain the fused time-series features, and input the fused time-series features into a multi-dimensional fault detection model to obtain the corresponding predicted fault detection result; Using time-series data and a large model for fault detection can not only determine the corresponding fault type when the disconnector has a fault, but also determine the possible fault types that the disconnector may generate in the future according to the time-series data of the disconnector within the preset time period. Compared with the prior art, the method of using a multi-dimensional fault detection model for fault detection does not require comparison with a large amount of data in the feature library, and at the same time does not require manual inspection to determine the fault type, thereby achieving the technical effect of improving the efficiency of fault detection of disconnectors.

[0081] The following uses specific embodiments to elaborate in detail on the technical solution of this application and how the technical solution of this application solves the above technical problems. The following 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 this application will be described below in conjunction with the accompanying drawings.

[0082] Figure 1 Flow schematic of the detection method for mechanical faults of the disconnector provided by this application Figure 1 , such as Figure 1 shown, this method includes:

[0083] S101. Collect time-series data of multiple dimensions generated by the disconnector within a preset time period.

[0084] In this step, the time-series data of multiple dimensions includes: mechanical vibration data, current data, and ambient temperature data of the disconnector.

[0085] It should be noted that the time-series data of multiple dimensions collected for the disconnector also includes: mechanical action time-series data of the disconnector, electrical drive time-series data of the disconnector, temperature time-series data of the disconnector, and external environment time-series data of the disconnector. Among them, the mechanical action time-series data can be: displacement time-series data, angle change time-series data, vibration time-series data, and sound change time-series data of the disconnector; the electrical drive time-series data can be: motor current change time-series data and voltage change time-series data of the disconnector; the temperature time-series data can be: contact temperature change time-series data and ambient temperature change time-series data of the disconnector; the external environment time-series data can be: wind speed change time-series data and humidity change time-series data.

[0086] In this step, the method for obtaining the time-series data of multiple dimensions of the disconnector within a preset time period can be: using multi-source sensors pre-deployed on the disconnector to collect the values at each moment within the preset time period, and integrating the values at each moment within the preset time period to obtain the time-series data of multiple dimensions.

[0087] Among them, the time-series data of multiple dimensions is used to indicate the change process of the operating state of the disconnector within a preset time period.

[0088] S102. Extract features for the time-series data of each dimension, and fuse the features of the time-series data of multiple dimensions to obtain fused time-series features.

[0089] In this step, when extracting features from the time series data for each dimension, it is necessary to extract the time domain features and frequency domain features of the time series data to generate a two-dimensional feature matrix corresponding to the time series data of each dimension; align and splice the two-dimensional feature matrices of each time series data to obtain the fused time series features corresponding to the time series data of multiple dimensions.

[0090] It should be noted that the generation of the fused time series features will be further explained in detail in the embodiments shown below, and no redundant elaboration will be made here. Figure 3 as shown below, and no redundant elaboration will be made here.

[0091] S103. Input the fused time series features into a multi-dimensional fault detection model to obtain the predicted fault detection result corresponding to the disconnector.

[0092] In this step, the multi-dimensional fault detection model is a multi-dimensional fault detection model obtained by training a time series prediction model based on the feature data of the disconnector in multiple dimensions.

[0093] Among them, the predicted fault detection result output by the multi-dimensional fault detection model includes the predicted fault type and the predicted fault occurrence time for the fused time series features; among them, the preset fault types are divided into two categories: the current fault type of the disconnector and the possible fault types that the disconnector may generate. The multi-dimensional fault detection model is mainly used to detect whether there is a current fault in the disconnector. If there is a fault, determine the corresponding fault type; if there is no current fault, predict the possible fault types of the disconnector according to the time series data.

[0094] Optionally, after obtaining the predicted fault result, a corresponding fault solution can also be generated for the predicted fault result. A possible implementation method for generating the fault solution is:

[0095] S1031. Match the predicted fault type and the predicted fault location in the predicted fault detection result with the preset knowledge base. When the match is successful, extract the disposal strategy corresponding to the predicted fault detection result from the preset knowledge base.

[0096] In this step, the preset knowledge base contains the processing strategies corresponding to different predicted fault types and different predicted fault locations of the disconnector.

[0097] Exemplarily, when performing fault analysis on the disconnector, in the predicted fault detection result output by the multi-dimensional fault detection model, the predicted fault type is: "poor contact"; the predicted fault location is: "main contact of the disconnector". The processing strategy found from the preset knowledge base according to the predicted fault type and the predicted fault location is: "Check and clean the contact point, and replace it if necessary".

[0098] S1032. Determine the priority corresponding to the predicted fault detection result based on the predicted fault occurrence probability and the predicted fault occurrence time in the predicted fault detection result.

[0099] In this step, when performing mechanical fault detection on the disconnector, after determining the processing strategy for each fault point according to the predicted fault type and the predicted fault location, it is necessary to determine the priority of each predicted fault detection result based on the predicted fault occurrence probability and the predicted fault occurrence time; thereby determining the priority of each processing strategy.

[0100] Optionally, a possible implementation of determining the priority is as follows:

[0101] a1. Calculate the priority score of the predicted fault detection result based on the predicted fault occurrence probability and the predicted fault occurrence time.

[0102] a2. Determine the priority of the predicted fault detection result based on the priority score.

[0103] S1033. Generate a fault solution corresponding to the predicted fault detection result based on the disposal strategy and the priority.

[0104] In this step, the fault solution includes the predicted fault detection results corresponding to the time series data within a continuous plurality of preset time periods, and the processing strategies corresponding to each predicted fault detection result.

[0105] Exemplarily, the length of the preset time period is 5 seconds, and the time difference between each preset time period is 1 second; the time series data of three preset time periods are collected; these three preset time periods are respectively: 0 second to 5 seconds, 1 second to 6 seconds, and 2 seconds to 7 seconds; the results obtained by performing fault detection on the time series data of these three preset time periods are respectively: the first predicted fault detection result, the second predicted fault detection result, and the third predicted fault detection result. Determine the processing strategies for each predicted fault detection result based on the predicted fault type and the predicted fault location of each predicted fault detection result, which are respectively: the first processing strategy, the second processing strategy, and the third processing strategy. Calculate the priority of each processing strategy according to the predicted fault occurrence probability and the predicted fault occurrence time of each predicted fault detection result; sort the processing strategies in descending order according to the priority, and generate a fault solution based on the sorted processing strategies.

[0106] The detection method for mechanical faults of the disconnector provided by the embodiments of the present application collects multi-dimensional time-series data generated by the disconnector within a preset time period, and obtains mechanical vibration data, current data, and ambient temperature data generated during the operation of the disconnector; performs feature extraction on the time-series data of each dimension, fuses the features of the time-series data of multiple dimensions to obtain fused time-series features; the fused time-series features contain the time-series data features of the disconnector in multiple dimensions; inputs the fused time-series features into a pre-trained multi-dimensional fault detection model to obtain the predicted fault detection result of the multi-dimensional fault detection model for the fused time-series features; thereby determines whether the disconnector has mechanical faults and the types of mechanical faults according to the preset fault detection results. Compared with the method of comparing with known fault cases in the prior art, the present application uses a multi-dimensional fault detection model to predict faults of the disconnector, which can not only analyze the existing faults of the disconnector, but also predict the possible faults of the disconnector, thereby achieving the technical effect of improving the efficiency of disconnector fault detection.

[0107] Figure 2 It is a schematic flowchart of the training method for the multi-dimensional fault detection model provided by the present application. As Figure 2 shown, based on the embodiment shown in Figure 1 this embodiment, the training of the multi-dimensional fault detection model in S103 is described in detail. The method includes:

[0108] S201. Perform data processing on the historical fault detection data to obtain a sample training data set and a sample validation data set.

[0109] In this step, the historical fault detection data includes the normal operation data and fault data of the disconnector.

[0110] Exemplarily, the historical fault detection data includes: the current, voltage, temperature, and vibration time-series data of the disconnector, as well as the fault label, fault location, and fault generation time.

[0111] Optionally, the data processing method can be: missing value removal, outlier removal, feature extraction, label assignment; and the data set is divided according to a preset ratio to obtain a sample training data set and a sample validation data set.

[0112] S202. Perform model training on the preset initial model based on the sample training data set to obtain a model to be determined.

[0113] In this step, the preset initial model refers to a time-series data prediction model used to output the prediction result corresponding to the time-series data based on the input time-series data.

[0114] Optionally, a possible implementation manner of the model training is:

[0115] S2021. Determine the layer structure, activation function, and loss function of the model based on a preset initial model and a preset model architecture.

[0116] S2022. Load the sample training data set and convert the format of the sample training data set into the format corresponding to the preset initial model.

[0117] S2023. Input the sample training data set into the preset initial model and perform iterative training on the preset initial model based on the layer structure, activation function, and loss function of the preset initial model.

[0118] S2024. When the loss value calculated by the loss function of the preset initial model is lower than the preset loss value, determine the trained preset initial model as the model to be determined.

[0119] S203. Perform anomaly verification on the model to be determined based on the sample verification data set. When the anomaly verification result of the model to be determined passes the verification, determine the model to be determined as a multi-dimensional fault detection model.

[0120] Optionally, a possible way to determine the multi-dimensional fault detection model using anomaly verification:

[0121] S2031. Calculate the anomaly score of the model to be determined based on the sample verification data set.

[0122] In this step, the anomaly score is used to indicate the probability of an abnormal prediction result in the model to be determined.

[0123] Optionally, a possible implementation of calculating the anomaly score is as follows:

[0124] b1. Input the sample verification data set into the model to be determined to obtain the output result of the model to be determined for each piece of data in the sample verification data set.

[0125] b2. Determine the predicted fault detection result corresponding to each piece of data according to each output result.

[0126] b3. Calculate the anomaly score based on the proportion of abnormal prediction results in the predicted fault detection results.

[0127] S2032. When the anomaly score belongs to the preset anomaly threshold range, determine the model to be determined as a multi-dimensional fault detection model.

[0128] Exemplarily, the preset anomaly threshold range is [0, 0.001]. When the calculated anomaly score belongs to this range, accept the model to be determined and determine it as a multi-dimensional fault detection model.

[0129] S2033. When the anomaly score does not fall within the preset anomaly threshold range, update the model to be determined based on the anomaly data in the sample verification dataset and their corresponding anomaly prediction results until the anomaly score of the model to be determined falls within the preset anomaly threshold range.

[0130] In this step, the method of updating the model to be determined is: adding the anomaly data in the dataset and their corresponding anomaly prediction results to the sample training dataset to retrain the model, and obtaining the updated model to be determined.

[0131] In this embodiment, the training and update of the multi-dimensional fault detection model are realized by using the anomaly score verification method, so as to improve the accuracy of model prediction.

[0132] Figure 3 It is a schematic flow of the detection of the mechanical fault of the disconnector provided by this application. Figure 2 , such as Figure 3 shown, based on the embodiments of Figure 1 and Figure 2 , this embodiment will elaborate on the detection of the mechanical fault of the disconnector in detail. The method includes:

[0133] S301. Connect the multi-source sensors pre-deployed on the disconnector.

[0134] In this step, the multi-source sensors include: current sensors, voltage sensors, temperature sensors, and vibration sensors.

[0135] It should be noted that this embodiment does not limit the specific quantity and type of the multi-source sensors. In actual situations, different types and unequal quantities of sensors can be deployed according to the scenario requirements to collect time-series data.

[0136] S302. For different types of time-series data, record the numerical changes of the multi-source sensors within a preset time period to obtain time-series data in multiple dimensions.

[0137] Optionally, a possible implementation method for collecting time-series data in multiple dimensions is:

[0138] S3021. Connect the multi-source sensors based on the preset interface parameters.

[0139] S3022. Based on the preset acquisition frequency and preset time period, collect the numerical changes of the multi-source sensors within the preset time period to obtain time-series data in multiple dimensions.

[0140] Among them, the preset acquisition frequency is the frequency of obtaining multi-source sensor values within a preset time period. Exemplarily, the duration of the preset time period is 10 seconds, and the preset acquisition frequency is 10 times per second. Then, there are 100 values for the time series data of each dimension collected, and these 100 values can represent the value change process of the disconnector in different dimensions within the preset time period.

[0141] S303. Perform data processing on the time series data of each dimension to obtain time series data of multiple dimensions with time alignment.

[0142] In this step, during the multi-source sensor data acquisition process, due to reasons such as the sampling frequencies, start times, or communication delays of different sensors, there is a problem that the time series data of multiple dimensions are not aligned in time. For subsequent analysis and modeling, it is necessary to perform alignment processing on the time series data of each dimension.

[0143] Optionally, a possible implementation method for its processing is as follows:

[0144] S3031. Convert the timestamps of the time series data of each dimension to the same time reference.

[0145] S3032. Determine the intersection range of the timestamps of the time series data of each dimension.

[0146] S3033. Select a target sampling frequency and resample all data to this frequency.

[0147] In this step, the resampling method can be: downsampling high-frequency data and performing interpolation processing on low-frequency data.

[0148] S3034. Align the time series data after resampling according to the timestamps to obtain the aligned time series data.

[0149] S304. Perform feature extraction on the time series data with time alignment to obtain the time series data features corresponding to the time series data of each dimension.

[0150] In this step, the time series data features include: time domain features, frequency domain features, and time-frequency domain features.

[0151] Among them, the time domain features include: mean, variance, peak value, kurtosis, skewness, zero-crossing rate; the frequency domain features include: spectral energy, main frequency, harmonic components, spectral entropy; the time-frequency domain features include: wavelet transform energy, short-time Fourier transform.

[0152] Integrate the time domain features, frequency domain features, and time-frequency domain features of each time series data to obtain the time series data features corresponding to the time series data of each dimension.

[0153] S305. Feature fusion is performed on the time series features corresponding to the time series data in multiple dimensions to obtain fused time series features.

[0154] In this step, the generation method of the fused features can be: splicing the time series features corresponding to the time series data of each dimension to obtain the fused time series features after splicing.

[0155] S306. Based on the attention mechanism, weighted processing is performed on the time series features of different dimensions in the fused time series features to obtain a joint feature matrix.

[0156] In this step, the attention mechanism is used to dynamically adjust the weights corresponding to the time series features of each dimension, for the key features of the book, and suppress redundant features.

[0157] Among them, the joint feature matrix refers to the feature matrix obtained by performing weighted processing on the data features of each dimension using the method of feature weighted calculation.

[0158] S307. Input the joint feature matrix into a multi-dimensional fault detection model to obtain a predicted fault detection result.

[0159] In this step, the predicted fault detection result includes a predicted fault type, a predicted fault location, a predicted fault occurrence probability, and a predicted fault occurrence time.

[0160] In this embodiment, through the acquisition of multi-dimensional time series data, the feature extraction and fusion of time series data, fused time series features are obtained; the multi-dimensional fault detection model is used to process the fused time series features to obtain a predicted fault detection result; by combining time series data and the model, the purpose of fault prediction based on time series data is achieved, thereby achieving the technical effect of improving the detection efficiency of disconnector mechanical faults.

[0161] Figure 4 It is a schematic structural diagram of the detection device for disconnector mechanical faults provided by this application, as Figure 4 shown, the detection device for disconnector mechanical faults provided by this embodiment includes:

[0162] An acquisition module 401, configured to acquire multi-dimensional time series data generated by the disconnector within a preset time period; among them, the multi-dimensional time series data includes: mechanical vibration data, current data, and ambient temperature data of the disconnector;

[0163] A first processing module 402, configured to perform feature extraction on the time series data of each dimension, and fuse the features of the multi-dimensional time series data to obtain fused time series features;

[0164] The second processing module 403 is configured to input the fused timing features into a multi-dimensional fault detection model to obtain a predicted fault detection result corresponding to the disconnector;

[0165] The multi-dimensional fault detection model is a multi-dimensional fault detection model obtained by training a time series prediction model based on feature data of the disconnector in multiple dimensions.

[0166] In a possible implementation, the apparatus further includes a third processing module 404, configured to:

[0167] Perform data processing on historical fault detection data to obtain a sample training data set and a sample verification data set;

[0168] Perform model training on a preset initial model based on the sample training data set to obtain a model to be determined;

[0169] Perform anomaly verification on the model to be determined based on the sample verification data set. When the anomaly verification result of the model to be determined passes the verification, determine the model to be determined as the multi-dimensional fault detection model.

[0170] In a possible implementation, the third processing module 404 is further configured to:

[0171] Calculate an anomaly score of the model to be determined based on the sample verification data set; the anomaly score is used to indicate the probability of an abnormal prediction result in the model to be determined;

[0172] When the anomaly score belongs to a preset anomaly threshold range, determine the model to be determined as the multi-dimensional fault detection model;

[0173] When the anomaly score does not belong to the preset anomaly threshold range, update the model to be determined based on the abnormal data and its corresponding abnormal prediction result in the sample verification data set until the anomaly score of the model to be determined belongs to the preset anomaly threshold range.

[0174] In a possible implementation, the acquisition module 401 is further configured to:

[0175] Connect to multi-source sensors pre-deployed on the disconnector;

[0176] For different types of timing data, record the numerical changes of the multi-source sensors within a preset time period to obtain timing data in multiple dimensions.

[0177] In a possible implementation, the first processing module 402 is further configured to:

[0178] Perform data processing on the timing data in each dimension to obtain timing data in multiple dimensions with timing alignment;

[0179] Extract features from time-series data for time-series alignment to obtain time-series data features corresponding to the time-series data in each dimension;

[0180] Fuse the time-series features corresponding to the time-series data in multiple dimensions to obtain fused time-series features.

[0181] In a possible implementation manner, the first processing module 402 is further configured to:

[0182] Perform weighted processing on the time-series features in different dimensions in the fused time-series features based on the attention mechanism to obtain a joint feature matrix;

[0183] Input the joint feature matrix into a multi-dimensional fault detection model to obtain a predicted fault detection result;

[0184] Wherein, the predicted fault detection result includes a predicted fault type, a predicted fault location, a predicted fault occurrence probability, and a predicted fault occurrence time.

[0185] In a possible implementation manner, the second processing module 403 is further configured to:

[0186] Match the predicted fault type and the predicted fault location in the predicted fault detection result with a preset knowledge base, and when the match is successful, extract a disposal strategy corresponding to the predicted fault detection result from the preset knowledge base;

[0187] Determine the priority corresponding to the predicted fault detection result based on the predicted fault occurrence probability and the predicted fault occurrence time in the predicted fault detection result;

[0188] Generate a fault solution corresponding to the predicted fault detection result based on the disposal strategy and the priority;

[0189] Wherein, the preset knowledge base includes different predicted fault types of the disconnector and processing strategies corresponding to different predicted fault locations.

[0190] The detection of the mechanical fault of the disconnector provided in this embodiment can execute the method provided in the above method embodiment, and its implementation principle and technical effect are similar, and will not be elaborated here in this embodiment.

[0191] Figure 5 It is a schematic structural diagram of an electronic device provided in this application. As Figure 5 shown, the electronic device provided in this embodiment includes: at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. Wherein, the processor 501, the memory 502, and the communication component 503 are connected through a bus 504.

[0192] In a specific implementation process, at least one processor 501 executes computer-executable instructions stored in a memory 502, so that at least one processor 501 executes the above-mentioned method.

[0193] For the specific implementation process of the processor 501, reference may be made to the above method embodiments. Their implementation principles and technical effects are similar, and will not be elaborated here in this embodiment.

[0194] In the above embodiments, it should be understood that the processor may be a central processing unit (Central Processing Unit, CPU for short), or may also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP for short), application specific integrated circuits (Application Specific Integrated Circuit, ASIC for short), 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 a hardware processor, or implemented by a combination of hardware and software modules in the processor.

[0195] 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.

[0196] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of this application is not limited to only one bus or one type of bus.

[0197] This application also provides a computer program product, including a computer program, which implements the above-mentioned method when executed by a processor.

[0198] 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-mentioned method is implemented.

[0199] The above-readable storage medium may 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 may be any available medium accessible by a general-purpose or special-purpose computer.

[0200] An exemplary readable storage medium is coupled to the processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium may also be an integral part of the processor. The processor and the readable storage medium may be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium may also exist as discrete components in a device.

[0201] The division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other may be indirect couplings or communication connections through some interfaces, devices, or units, and may be in electrical, mechanical, or other forms.

[0202] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0203] In addition, in each embodiment of the present invention, the functional units may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit.

[0204] If a 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 this 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 may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments 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.

[0205] 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.

[0206] Finally, it should be noted that: After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily think of other implementation manners of the present invention. The present invention is intended to cover any variations, uses, or adaptive changes of the present invention. These variations, uses, or adaptive changes follow the general principles of the present invention and include the common general knowledge or conventional technical means in the technical field of the present invention that are 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 method for detecting a mechanical failure of an isolating switch, characterized in that: include: Collecting time series data of multiple dimensions generated by the isolating switch within a preset time period; wherein the time series data of multiple dimensions include: mechanical vibration data, current data, and ambient temperature data of the isolating switch; Extract features from the time series data of each dimension, and fuse the features of the time series data of multiple dimensions to obtain fused time series features; Inputting the fused time series features into a multi-dimensional fault detection model to obtain a predicted fault detection result corresponding to the disconnector; The multi-dimensional fault detection model is obtained by training a time series prediction model based on feature data of the isolating switch in multiple dimensions.

2. The method according to claim 1, characterized in that Before collecting the time series data of multiple dimensions generated by the isolating switch within a preset time period, the method further includes: Perform data processing based on historical fault detection data to obtain a sample training data set and a sample verification data set; Performing model training on a preset initial model based on the sample training data set to obtain a model to be determined; An abnormality check is performed on the model to be determined based on the sample verification data set, and when the abnormality check result of the model to be determined is that the check passes, the model to be determined is determined as a multi-dimensional fault detection model.

3. The method according to claim 2, characterized in that The performing an outlier check on the model to be determined based on the sample verification set, and determining the model to be determined as a multi-dimensional fault detection model when the outlier check result of the model to be determined is that the check passes, includes: Calculating an anomaly score of the model to be determined based on the sample verification data set; wherein the anomaly score is used to indicate the probability of an abnormal prediction result occurring in the model to be determined; When the abnormality score belongs to the preset abnormality threshold range, determining the model to be determined as a multi-dimensional fault detection model; When the anomaly score does not fall within the preset anomaly threshold range, the model to be determined is updated based on the anomaly data in the sample verification data set and its corresponding anomaly prediction result until the anomaly score of the model to be determined falls within the preset anomaly threshold range.

4. The method according to claim 1, characterized in that: The collecting of time series data of multiple dimensions generated by the isolating switch within a preset time period includes: Connect multi-source sensors pre-deployed on disconnect switches; For different types of time series data, the value changes of the multi-source sensors within the preset time period are recorded to obtain time series data of multiple dimensions.

5. The method according to claim 1 or 4, characterized in that: The feature extraction is performed on the time series data of each dimension, and the features corresponding to the time series data of multiple dimensions are fused to obtain the fused time series features, including: Perform data processing on the time series data of each dimension to obtain time series data of multiple dimensions with time alignment; Perform feature extraction on the time series data with time series alignment to obtain time series data features corresponding to the time series data in each dimension; The time series features corresponding to the time series data of multiple dimensions are fused to obtain fused time series features.

6. The method according to claim 5, characterized in that The step of inputting the fused time series features into a multi-dimensional fault detection model to obtain a predicted fault detection result corresponding to the disconnector includes: Based on the attention mechanism, weighted processing is performed on the time series features of different dimensions in the fused time series features to obtain a joint feature matrix; Inputting the joint feature matrix into the multi-dimensional fault detection model to obtain the predicted fault detection result; The predicted fault detection result includes the predicted fault type, predicted fault location, predicted fault occurrence probability, and predicted fault occurrence time.

7. The method according to claim 6, characterized in that After inputting the fused time series features into the multi-dimensional fault detection model to obtain the predicted fault detection result corresponding to the disconnector, the method further includes: Matching the predicted fault type and predicted fault location in the predicted fault detection result with a preset knowledge base, and when the match is successful, extracting a disposal strategy corresponding to the predicted fault detection result from the preset knowledge base; Determine the priority corresponding to the predicted fault detection result based on the predicted fault occurrence probability and the predicted fault occurrence time in the predicted fault detection result; Based on the handling strategy and the priority, generating a fault solution corresponding to the predicted fault detection result; The preset knowledge base includes different predicted fault types of the disconnector and corresponding processing strategies for different predicted fault locations.

8. A detection device for mechanical failure of an isolating switch, characterized in that: include: An acquisition module, used to collect time series data of multiple dimensions generated by the isolating switch within a preset time period; wherein the time series data of multiple dimensions include: mechanical vibration data, current data, and ambient temperature data of the isolating switch; The first processing module is used to extract features from the time series data of each dimension, and fuse the features of the time series data of multiple dimensions to obtain fused time series features; A second processing module is used to input the fused time series features into a multi-dimensional fault detection model to obtain a predicted fault detection result corresponding to the disconnector; The multi-dimensional fault detection model is obtained by training a time series prediction model based on feature data of the isolating switch in multiple dimensions.

9. An electronic device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

11. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images