Cable state monitoring and fault diagnosis device, method, equipment and medium

Abstract

The invention discloses a cable state monitoring and fault diagnosis device and method, equipment and a medium, and the device comprises a data collection module and an intelligent diagnosis unit, and the data collection module collects multi-source data used for diagnosing the state of a cable; the method comprises the following steps: converting multi-source data into a space-time matrix through an intelligent diagnosis unit, carrying out feature extraction, outputting a fault type and a corresponding prediction probability of a cable by adopting a preset multi-label classification and dynamic threshold decision-making mechanism according to the extracted features, and sending out early warning signals of different grades according to the prediction probability. The diagnosis device solves the problems that in the prior art, multi-dimensional data are not fully fused, so that the diagnosis result is not comprehensive and accurate enough; and new fault features caused by cable aging or operation environment change cannot be self-adapted, the model generalization ability is limited, and the model adaptability is poor.

Description

Technical Field

[0001] The present invention relates to the technical field of cable condition monitoring, and in particular to a cable condition monitoring and fault diagnosis device, method, equipment and medium. Background Art

[0002] In the technical field of power equipment monitoring, as an important carrier for transmitting electric energy and signals, the operating state of cables is directly related to the safety and stability of the power system. However, due to factors such as complex cable laying environments, limited service lives, and external force damage, cable faults occur frequently, causing inconvenience to production and life and even significant economic losses. Therefore, cable condition monitoring and fault diagnosis technologies are particularly important.

[0003] Traditional cable monitoring systems mainly use methods such as manual analysis or simple threshold judgment, which have problems such as relying on manual analysis, insufficient real-time performance, low data utilization rate, and poor adaptability. For example, traditional methods such as the fault locator method, thermal infrared detection method, and high-frequency pulse detection method can, to a certain extent, detect and locate cable faults, but it is difficult to handle complex fault modes with multi-parameter coupling, and it is impossible to adapt to new fault characteristics brought about by cable aging or changes in the operating environment.

[0004] In the prior art, there are already some cable fault diagnosis methods based on deep learning. For example, convolutional neural networks are used to classify and identify cable fault images, or a combination of various sensor data is used to monitor and warn of the cable condition. However, these methods do not fully integrate multi-dimensional data such as leakage current, temperature and humidity, and temperature field distribution, resulting in incomplete and inaccurate diagnostic results; and they cannot adapt to new fault characteristics brought about by cable aging or changes in the operating environment, and the model generalization ability is limited, resulting in poor model adaptability. Summary of the Invention

[0005] The present invention provides a cable condition monitoring and fault diagnosis device, method, equipment and medium, which are used to solve the problems in the prior art that multi-dimensional data are not fully integrated, resulting in incomplete and inaccurate diagnostic results; and it is impossible to adapt to new fault characteristics brought about by cable aging or changes in the operating environment, and the model generalization ability is limited, resulting in poor model adaptability.

[0006] In view of this, in the first aspect of the present invention, a cable condition monitoring and fault diagnosis device is provided. The device includes: a data acquisition module, a dual-mode communication module, and an intelligent diagnosis unit; wherein, the data acquisition module includes several information acquisition sensors of different types; the intelligent diagnosis unit includes: a human-computer interaction module and a CNN model composed of an input layer, a feature extraction layer, and a decision layer.

[0007] The data acquisition module is used to collect several different types of data for diagnosing the cable status to obtain multi-source data after responding to the diagnosis instruction sent by the human-computer interaction module, and send the multi-source data to the intelligent diagnosis unit through the dual-mode communication module;

[0008] The intelligent diagnosis unit is used to convert the multi-source data into a spatio-temporal matrix through the input layer and input it into the feature extraction layer, capture the spatio-temporal correlation features of the spatio-temporal matrix by using a 3D convolution kernel through the feature extraction layer and input it into the decision layer, output the fault type of the cable and the corresponding prediction probability by using a preset multi-label classification and dynamic threshold decision mechanism through the decision layer, and send out warning signals of different levels through the human-computer interaction module according to the prediction probability.

[0009] Optionally, the converting the multi-source data into a spatio-temporal matrix through the input layer and inputting it into the feature extraction layer includes:

[0010] Performing data standardization processing on the multi-source data;

[0011] Splicing the standardized multi-source data according to the time dimension to form a four-dimensional spatio-temporal matrix;

[0012] Taking the four-dimensional spatio-temporal matrix as training data, performing random time window truncation, small-scale spatial translation and contrast adjustment processing on the training data, and then inputting it into the feature extraction layer.

[0013] Optionally, when the multi-source data includes: leakage current, ambient humidity around the cable, temperature of the cable joint and outer shell, and cable arc light, the performing data standardization processing on the multi-source data includes:

[0014] Mapping the original current value of the leakage current to the interval [0, 1] through Min-Max normalization, and retaining the local abnormal pulse characteristics;

[0015] Performing standardization processing on the ambient humidity around the cable by using Z-Score standardization;

[0016] Performing standardization processing on the temperature of the cable joint and outer shell by using the logarithmic compression method;

[0017] Performing standardization processing on the cable arc light by combining grayscale conversion and histogram equalization.

[0018] Optionally, the capturing the spatio-temporal correlation features of the spatio-temporal matrix by using a 3D convolution kernel through the feature extraction layer includes:

[0019] Using a 3×3×3 convolution kernel and designing cross-channel shared convolution kernel parameters to capture the correlation features of both the time dimension and the space dimension of the spatio-temporal matrix simultaneously.

[0020] Optionally, the design method of the preset multi-label classification and dynamic threshold decision mechanism includes:

[0021] Design the decision layer to consist of a fully connected layer with a Sigmoid activation function. The output dimension is equal to the predefined number of fault types. Each output node corresponds to the independent prediction probability of a fault type. Use binary cross-entropy loss as the loss function and design the class weights of the fault types to obtain the weighted loss function. Adopt a dynamic adjustment strategy based on the fault alarm frequency within the sliding window as the early warning threshold.

[0022] Optionally, the expression of the weighted loss function is:

[0023] ;

[0024] where N is the number of samples, is the true label of the c-th type of fault for the i-th sample, is the predicted probability, is the weight coefficient of the c-th class.

[0025] Optionally, the expression of the early warning threshold is:

[0026] ;

[0027] where is the basic threshold, is the adjustment coefficient, is the length of the sliding window, is the alarm status of the c-th type of fault at time

[0028] The second aspect of the present invention provides a cable status monitoring and fault diagnosis method, and the method includes:

[0029] After responding to the diagnosis instruction, collect several different types of data for diagnosing the cable status to obtain multi-source data;

[0030] Convert the multi-source data into a spatio-temporal matrix, and use a 3D convolution kernel to capture the spatio-temporal correlation features of the spatio-temporal matrix,

[0031] Use the preset multi-label classification and dynamic threshold decision mechanism to process the spatio-temporal correlation features to obtain the fault type of the cable and the corresponding prediction probability;

[0032] Send out early warning signals of different levels according to the prediction probability.

[0033] The third aspect of the present invention provides a cable status monitoring and fault diagnosis device, and the device includes a processor and a memory:

[0034] The memory is used to store program codes and transmit the program codes to the processor;

[0035] The processor is used to execute the steps of the cable status monitoring and fault diagnosis method described in the second aspect above according to the instructions in the program codes.

[0036] A fourth aspect of the present invention provides a computer-readable storage medium, which is used to store program codes for executing the method described in the second aspect above.

[0037] It can be seen from the above technical solutions that the present invention has the following advantages:

[0038] The embodiment of the present invention provides a cable status monitoring and fault diagnosis device. First, the data acquisition module acquires various different types of data for diagnosing the cable status, thereby obtaining multi-source data. Subsequently, the intelligent diagnosis unit converts these multi-source data into a spatio-temporal matrix. By fusing multi-source heterogeneous data and extracting spatio-temporal features, the cable status can be more comprehensively understood, thereby improving the accuracy of fault diagnosis. Then, the decision-making layer of the CNN model uses a preset multi-label classification and dynamic threshold decision-making mechanism to output the fault types of the cable and their corresponding prediction probabilities, and issue early warning signals of different levels according to these probabilities. It can be understood that through the CNN acceleration engine, automatic extraction of fault features and real-time diagnosis can be achieved, meeting the strict requirements of the power system for real-time performance. More importantly, through incremental learning, the model can be adaptively updated to adapt to new fault features caused by cable aging or changes in the operating environment, thereby enhancing the generalization ability of the model. Thus, the problems in the prior art that multi-dimensional data are not fully fused, resulting in incomplete and inaccurate diagnostic results; and the model cannot adapt to new fault features brought about by cable aging or changes in the operating environment, and the generalization ability of the model is limited, resulting in poor adaptability of the model are solved. Description of the Drawings

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0040] Figure 1 It is a schematic structural diagram of a cable status monitoring and fault diagnosis device provided by an embodiment of the present invention;

[0041] Figure 2Schematic diagram of a cable condition monitoring and fault diagnosis device provided by an embodiment of the present invention. Detailed implementation manners

[0042] To make the objectives, features, and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described below are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0043] Please refer to Figure 1 , a cable condition monitoring and fault diagnosis device provided in an embodiment of the present invention includes: a data acquisition module, a dual-mode communication module, and an intelligent diagnosis unit; wherein, the data acquisition module includes several information acquisition sensors of different types; the intelligent diagnosis unit includes: a human-computer interaction module and a CNN model composed of an input layer, a feature extraction layer, and a decision layer.

[0044] It should be noted that the CNN model, that is, the Convolutional Neural Network model, is a deep learning model specifically designed for processing data with grid structures (such as two-dimensional pixel grids in images and one-dimensional time series in audio).

[0045] The data acquisition module is used to collect several different types of data for diagnosing the cable condition to obtain multi-source data after responding to the diagnosis instruction sent by the human-computer interaction module, and send it to the intelligent diagnosis unit through the dual-mode communication module.

[0046] It should be noted that the multi-source data includes but is not limited to cable temperature data, current data, voltage data, and environmental factor data, etc. These data can comprehensively reflect the operating state of the cable. The data acquisition module obtains these data in real time through high-precision sensors to ensure the accuracy and reliability of the data. At the same time, the dual-mode communication module uses both wireless and wired communication methods to ensure the stability and flexibility of data transmission, and can ensure the timely transmission of data even in complex environments.

[0047] The intelligent diagnosis unit is used to convert the multi-source data into a spatio-temporal matrix through the input layer and input it into the feature extraction layer, capture the spatio-temporal correlation features of the spatio-temporal matrix through the feature extraction layer using a 3D convolution kernel and input it into the decision layer, output the cable fault type and the corresponding prediction probability through the decision layer using a preset multi-label classification and dynamic threshold decision mechanism, and issue early warning signals of different levels through the human-computer interaction module according to the prediction probability.

[0048] It should be noted that the intelligent diagnostic unit is the core component of the device, integrating advanced deep learning algorithms, particularly for processing spatiotemporal data. At the input layer, preprocessed multi-source data forms a four-dimensional spatiotemporal matrix, which is efficiently received and laid the foundation for subsequent feature extraction. In the feature extraction layer, the application of 3D convolution kernels is crucial, as they can deeply explore the temporal correlations and spatial features hidden in the spatiotemporal matrix, which are crucial for accurately determining cable status. The decision layer further enhances diagnostic accuracy. Using multi-label classification technology, the system identifies various possible cable fault types and, combined with a dynamic threshold decision mechanism, assigns a predicted probability to each fault type. This mechanism not only improves diagnostic sensitivity but also ensures the accuracy and relevance of warning signals. When the predicted probability exceeds a preset threshold, the intelligent diagnostic unit immediately issues a warning signal of the corresponding level through the human-computer interaction module, enabling operators to quickly respond and take effective measures.

[0049] In one embodiment, the multi-source data to be collected include: leakage current, ambient humidity of the cable, cable joint and casing temperature, and cable arcing;

[0050] like Figure 1 As shown, in a specific implementation, the data acquisition module includes: a leakage current measurement module, an ambient humidity module, a connector / casing temperature module and an arc detection module; these modules respectively collect leakage current, ambient humidity around the cable, cable connector and casing temperature, and cable arc; further, the data acquisition module is also provided with a power supply module; a dual-mode communication module: supports local LoRa and remote 4G / 5G communications to realize real-time data transmission and remote monitoring.

[0051] in:

[0052] Power supply module: adopts low-power and wide-voltage power supply design to power each acquisition module, ensuring stable operation of each acquisition module in different environments.

[0053] Leakage current measurement module: Utilizes high-frequency CT sensors to achieve high-precision measurement of leakage current (accuracy ±0.5mA).

[0054] Ambient humidity module: uses a capacitive sensor with a measurement range of 0-100%RH to monitor the ambient humidity around the cable in real time.

[0055] Connector / Casing Temperature Module: Utilizes an infrared array sensor with a resolution of 0.1°C to monitor the cable connector and casing temperature in real time.

[0056] Arc detection module: uses a passive arc sensor to collect specific frequency bands of arc signals, filter out common spectrum interference, and the light sensitivity can be adjusted according to actual conditions to monitor cable arcs in real time.

[0057] Furthermore, in one embodiment, converting various types of data into a spatiotemporal matrix and inputting it into a feature extraction layer includes the following steps:

[0058] 1. Perform data standardization on multi-source data;

[0059] It should be noted that:

[0060] Leakage current: The original current value (unit: mA) is mapped to the [0, 1] interval through Min-Max normalization, preserving the local abnormal pulse characteristics;

[0061] ;

[0062] in, is the original current value, and are the minimum and maximum current values in the training data set, respectively. The output range is [0,1], and the original pulse waveform characteristics are retained.

[0063] Cable ambient humidity: The ambient temperature and humidity are standardized using Z-Score to eliminate measurement deviations from environmental sensors and highlight temperature and humidity change trends.

[0064] ;

[0065] in, and is the original temperature and humidity value, and is the mean temperature and humidity, and is the standard deviation of temperature and humidity.

[0066] Cable connector and housing temperature: Logarithmic compression is used to normalize the temperature field distribution to enhance the contrast of high-temperature areas. The temperature matrix (unit: °C) output by the infrared thermal imager is logarithmically compressed.

[0067] ;

[0068] in, is a small constant to prevent division by zero, and scale is the scaling factor. Take 10 -6 , is a small constant to prevent division by zero, and scale=100 is a scaling factor to enhance the dynamic range in the high temperature region.

[0069] Cable arc: Arc image binary mask generation enhances the brightness of the arc area through grayscale + histogram equalization, and generates an arc mask matrix after binarization.

[0070] Grayscale: ;

[0071] Histogram equalization uses the equalizeHist function, and the binarization formula is:

[0072] ;

[0073] 2. Concatenate the standardized multi-source data along the time dimension to form a four-dimensional spatio-temporal matrix;

[0074] [Time window length × height × width × number of channels]

[0075] It should be noted that in the time dimension, it is set to 10 consecutive sampling periods (for example, sampling once per second, and the time window covers 10 seconds). In the spatial dimension, the leakage current and temperature and humidity data are interpolated and extended to the same spatial resolution as the temperature field distribution (such as a 32×32 grid), and the arc image is directly used as an independent channel (single-channel binary matrix). The total number of channels = leakage current channel (1) + temperature and humidity channel (1) + temperature field channel (1) + arc channel (1) = 4 channels.

[0076] 3. Use the four-dimensional spatio-temporal matrix as training data, and after randomly intercepting the time window, slightly translating the space, and adjusting the contrast of the training data, input it into the feature extraction layer.

[0077] It should be noted that using the four-dimensional spatio-temporal matrix as training data and performing random time window interception, slight space translation, and contrast adjustment on the training data can improve the generalization ability of the model.

[0078] It should be noted that standardizing the multi-source data ensures the consistency of data from different sources in terms of numerical range and distribution characteristics, laying a solid foundation for the subsequent construction of the spatio-temporal matrix. Standardization not only helps to eliminate the dimensional differences between data, but also improves the stability and convergence speed of model training. Through this step, we can integrate diverse monitoring data into a unified framework, facilitating subsequent feature extraction and pattern recognition.

[0079] In one embodiment, the feature extraction layer uses a 3D convolutional kernel to capture the spatio-temporal correlation features of the spatio-temporal matrix, including:

[0080] Use a 3×3×3 convolutional kernel and design cross-channel shared convolutional kernel parameters to simultaneously capture the correlation features of the time dimension and the space dimension of the spatio-temporal matrix.

[0081] It should be noted that the feature extraction layer uses a 3×3×3 convolutional kernel to simultaneously capture the correlation features of the time dimension (fault evolution trend) and the space dimension (local state pattern), as shown in Table 1. Cross-channel sharing of convolutional kernel parameters reduces the number of parameters and forces the learning of a general feature representation.

[0082] Table 1 details the specific performance of using a 3×3×3 convolutional kernel in the feature extraction process, which not only shows the trend characteristics of fault evolution in the time dimension but also clearly depicts the pattern characteristics of local states in the space dimension. These characteristics together constitute a comprehensive and detailed description of the cable state, providing a rich information basis for subsequent diagnosis. The design of sharing convolutional kernel parameters across channels not only significantly reduces the number of model parameters, alleviates the computational burden, but also enhances the generalization ability and robustness of the model by forcing the learning of general feature representations. This innovative design idea enables our device to exhibit more stable and accurate monitoring and diagnosis performance when facing complex and variable cable states.

[0083] Table 1

[0084]

[0085] In one embodiment, the design method of the preset multi-label classification and dynamic threshold decision-making mechanism includes:

[0086] Design the decision layer to be composed of a fully connected layer with a Sigmoid activation function, where the output dimension is equal to the number of predefined fault types. Each output node corresponds to the independent prediction probability of a fault type. Use binary cross-entropy loss as the loss function and design the class weights of the fault types to obtain the weighted loss function. Adopt a dynamic adjustment strategy based on the fault alarm frequency within the sliding window as the warning threshold.

[0087] Among them, the expression of the weighted loss function is:

[0088] ;

[0089] In the formula, N is the number of samples, is the true label of the c-th type of fault for the i-th sample, is the predicted probability, is the weight coefficient of the c-th class.

[0090] The expression of the warning threshold is:

[0091] ;

[0092] In the formula, is the basic threshold, is the adjustment coefficient, is the length of the sliding window, is the alarm state of the c-th type of fault at the th moment.

[0093] It should be noted that the decision-making layer of this embodiment adopts a multi-label classification and dynamic threshold decision-making mechanism, and the specific design is as follows:

[0094] The decision-making layer is composed of a fully connected layer with a Sigmoid activation function, and the output dimension is equal to the predefined number of fault types (set to C types, for example, C = 4 corresponds to four types of faults: local overheating, insulation aging, arc discharge, and environmental anomaly). Each output node corresponds to the independent prediction probability of a fault type, and the value range is between (0, 1). The loss function adopts binary cross-entropy loss (BCE Loss), and the calculation formula is:

[0095] ;

[0096] where N is the number of samples, is the true label (0 or 1) of the c-th type of fault for the i-th sample, is the prediction probability.

[0097] To alleviate the problem of class imbalance, the class weight is set, and the calculation method is the total number of samples in the training set divided by the number of samples of the c-th class:

[0098] ;

[0099] The weighted loss function is modified to:

[0100] ;

[0101] The threshold is dynamically adjusted according to the fault alarm frequency within the sliding window ;

[0102] where is the basic threshold, is the adjustment coefficient (dynamically adjusted according to the historical false alarm rate), W is the length of the sliding window, is the alarm status (0 or 1) of the c-th type of fault at the k-th moment. When the prediction probability of any label exceeds the corresponding threshold, an alarm is triggered, and the top N types of faults with the highest probability are output as a reference:

[0103] If then an alarm for the c-th type of fault is triggered, and the reference fault type .

[0104] Regarding the above-mentioned warning signals of different levels sent by the human-machine interaction module according to the prediction probability, it can be understood that the fault type is output. When the prediction probability of any fault type first exceeds the basic threshold (0.6), a primary warning is triggered. When the prediction probability continuously exceeds the danger threshold (0.85) for 2 detection cycles or exceeds the emergency threshold (0.92) in a single cycle, a secondary warning is triggered. When a multi-label coupled fault is detected (such as overheating + arc discharge occurring simultaneously) or the prediction probability exceeds 0.98, a tertiary warning is initiated. The thresholds can be adjusted arbitrarily, and a touch screen and voice alarm function are provided to facilitate user operation and fault warning.

[0105] An embodiment of the present invention provides a cable status monitoring and fault diagnosis device. First, the data acquisition module acquires various different types of data for diagnosing the cable status, thereby obtaining multi-source data. Subsequently, the intelligent diagnosis unit converts this multi-source data into a spatio-temporal matrix. By fusing multi-source heterogeneous data and extracting spatio-temporal features, the cable status can be understood more comprehensively, thereby improving the accuracy of fault diagnosis. Then, the decision-making layer of the CNN model uses a preset multi-label classification and dynamic threshold decision mechanism to output the fault type of the cable and its corresponding prediction probability, and issues warning signals of different levels based on these probabilities. It can be understood that through the CNN acceleration engine, automatic extraction and real-time diagnosis of fault features can be achieved, meeting the strict requirements of the power system for real-time performance. More importantly, through incremental learning, the model can be adaptively updated to adapt to new fault features caused by cable aging or changes in the operating environment, thereby enhancing the generalization ability of the model. Thus, it solves the problems that the prior art does not fully fuse multi-dimensional data, resulting in incomplete and inaccurate diagnostic results; and it cannot adapt to new fault features brought about by cable aging or changes in the operating environment, and the generalization ability of the model is limited, resulting in poor adaptability of the model.

[0106] The above is a cable status monitoring and fault diagnosis device provided in an embodiment of the present invention. The following is a cable status monitoring and fault diagnosis method provided in an embodiment of the present invention.

[0107] Please refer to Figure 2 , a cable status monitoring and fault diagnosis method provided in an embodiment of the present invention includes:

[0108] Step 101: After responding to the diagnosis instruction, acquire several different types of data for diagnosing the cable status to obtain multi-source data.

[0109] Step 102: Convert the multi-source data into a spatio-temporal matrix, and use a 3D convolution kernel to capture the spatio-temporal correlation features of the spatio-temporal matrix.

[0110] Step 103: Process the spatio-temporal correlation features using a preset multi-label classification and dynamic threshold decision mechanism to obtain the fault type of the cable and the corresponding prediction probability.

[0111] Step 104: Send warning signals of different levels according to the predicted probability.

[0112] The embodiment of the present invention provides a method for cable condition monitoring and fault diagnosis. First, various different types of data for diagnosing the cable condition are collected to obtain multi-source data. Subsequently, these multi-source data are converted into a spatio-temporal matrix. By fusing multi-source heterogeneous data and extracting spatio-temporal features, the cable condition can be understood more comprehensively, thereby improving the accuracy of fault diagnosis. Then, using a preset multi-label classification and dynamic threshold decision mechanism, the fault type of the cable and its corresponding predicted probability are output, and warning signals of different levels are sent according to these probabilities. It can be understood that through the CNN acceleration engine, automatic extraction and real-time diagnosis of fault features can be achieved, meeting the strict requirements of the power system for real-time performance. More importantly, through incremental learning, the model can be adaptively updated to adapt to new fault features caused by cable aging or changes in the operating environment, thereby enhancing the generalization ability of the model. Thus, it solves the problems in the prior art that multi-dimensional data are not fully fused, resulting in incomplete and inaccurate diagnostic results; and it cannot adapt to new fault features brought about by cable aging or changes in the operating environment, and the generalization ability of the model is limited, resulting in poor adaptability of the model.

[0113] Furthermore, the embodiment of the present invention also provides a cable condition monitoring and fault diagnosis device, and the device includes a processor and a memory:

[0114] The memory is used to store program codes and transmit the program codes to the processor;

[0115] The processor is used to execute the steps of the cable condition monitoring and fault diagnosis method as described in the above method embodiment according to the instructions in the program codes.

[0116] Furthermore, the embodiment of the present invention also provides a computer-readable storage medium, and the computer-readable storage medium is used to store program codes, and the program codes are used to execute the cable condition monitoring and fault diagnosis method as described in the above method embodiment.

[0117] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0118] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there can 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 displayed or discussed coupling, direct coupling, or communication connection between each other can be through some interfaces, and the indirect coupling or communication connection of devices or units can be in electrical, mechanical, or other forms.

[0119] 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 they 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.

[0120] In addition, each functional unit in various embodiments of the present invention can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0121] If the integrated unit 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 all or 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 various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, and other media that can store program codes.

[0122] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of various embodiments of the present invention.

Claims

1. A cable condition monitoring and fault diagnosis device, characterized in that Including: A data acquisition module, a dual-mode communication module, and an intelligent diagnosis unit; wherein, the data acquisition module includes several information acquisition sensors of different types; the intelligent diagnosis unit includes: a human-computer interaction module and a CNN model composed of an input layer, a feature extraction layer, and a decision layer; The data acquisition module is configured to, after responding to a diagnosis instruction sent by the human-computer interaction module, acquire several different types of data for diagnosing the cable state to obtain multi-source data, and send the multi-source data to the intelligent diagnosis unit through the dual-mode communication module; The intelligent diagnosis unit is configured to convert the multi-source data into a spatio-temporal matrix through the input layer and input the spatio-temporal matrix into the feature extraction layer, capture the spatio-temporal correlation features of the spatio-temporal matrix through the feature extraction layer using a 3D convolution kernel and input the spatio-temporal correlation features into the decision layer, output the fault type and the corresponding prediction probability of the cable through the decision layer using a preset multi-label classification and dynamic threshold decision mechanism, and issue early warning signals of different levels through the human-computer interaction module according to the prediction probability.

2. The cable status monitoring and fault diagnosis device according to claim 1, characterized in that The converting the multi-source data into a spatio-temporal matrix through the input layer and inputting the spatio-temporal matrix into the feature extraction layer includes: Performing data normalization processing on the multi-source data; Splicing the normalized multi-source data according to the time dimension to form a four-dimensional spatio-temporal matrix; Using the four-dimensional spatio-temporal matrix as training data, performing random time window truncation, small-scale spatial translation, and contrast adjustment processing on the training data, and then inputting the processed training data into the feature extraction layer.

3. The cable status monitoring and fault diagnosis device according to claim 2, characterized in that When the multi-source data includes: leakage current, ambient humidity around the cable, temperature of the cable joint and the outer casing, and cable arc light, the performing data normalization processing on the multi-source data includes: Mapping the original current value of the leakage current to the interval [0, 1] through Min-Max normalization, and retaining the local abnormal pulse features; Performing standardization processing on the ambient humidity around the cable using Z-Score standardization; Performing standardization processing on the temperature of the cable joint and the outer casing using a logarithmic compression method; Combining grayscale conversion and histogram equalization to perform standardization processing on the cable arc light.

4. The cable status monitoring and fault diagnosis device according to claim 2, characterized in that, The capturing the spatio-temporal correlation features of the spatio-temporal matrix through the feature extraction layer using a 3D convolution kernel includes: Using a 3×3×3 convolution kernel and designing cross-channel shared convolution kernel parameters to capture the correlation features of the time dimension and the spatial dimension of the spatio-temporal matrix simultaneously.

5. The cable status monitoring and fault diagnosis device according to claim 1, characterized in that The design method of the preset multi-label classification and dynamic threshold decision mechanism includes: Designing the decision layer to be composed of a fully connected layer with a Sigmoid activation function, the output dimension being equal to the number of predefined fault types, each output node corresponding to an independent prediction probability of a fault type, using binary cross-entropy loss as the loss function, and designing the class weights of the fault types to obtain a weighted loss function, and using a dynamic adjustment strategy of the fault alarm frequency within a sliding window as the early warning threshold.

6. The cable status monitoring and fault diagnosis device according to claim 5, characterized in that, The expression of the weighted loss function is: ； where N is the number of samples, is the true label of the c-th type of fault for the i-th sample, is the predicted probability, is the weight coefficient for the c-th class.

7. The cable status monitoring and fault diagnosis device according to claim 5, characterized in that The expression of the early warning threshold is: ； In the formula, is the basic threshold, is the adjustment coefficient, is the sliding window length, is the alarm status of the c - type fault at the moment.

8. A method for cable status monitoring and fault diagnosis, characterized in that, Including: After responding to the diagnosis instruction, acquiring several different types of data for diagnosing the cable state to obtain multi-source data; Convert the multi-source data into a spatio-temporal matrix, and use a 3D convolution kernel to capture the spatio-temporal correlation features of the spatio-temporal matrix; Process the spatio-temporal correlation features by using a pre-set multi-label classification and dynamic threshold decision mechanism to obtain the fault type of the cable and the corresponding prediction probability; Send warning signals of different levels according to the prediction probability.

9. A cable condition monitoring and fault diagnosis device, characterized in that The device includes a processor and a memory: The memory is used to store program codes and transmit the program codes to the processor; The processor is used to execute the cable status monitoring and fault diagnosis method according to the instructions in the program codes as claimed in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program codes, and the program codes are used to execute the cable status monitoring and fault diagnosis method as claimed in claim 8.

Images