Multi-fusion cable state analysis method and device, terminal equipment and storage medium

Abstract

The invention discloses a multi-fusion-based cable state analysis method and device, terminal equipment and a storage medium, and belongs to the technical field of cable fault diagnosis, and the method comprises the steps: obtaining cable operation time sequence data; wherein the cable operation time sequence data comprises a cable partial discharge time sequence signal, cable dielectric loss time sequence data, cable circulation time sequence data, cable temperature time sequence data, environment temperature time sequence data and cable load current time sequence data; updating a fusion weight vector of the DSFF module according to the cable operation time sequence data; fusing the cable operation feature data according to the updated fusion weight vector to obtain fusion feature data; and predicting the cable state according to the fusion. Due to the fact that the cable operation time sequence data obtained in real time comprises the data related to the cable operation condition and the data related to the environment condition, the problem that the prior art is difficult to adapt to changes of the cable operation condition and the environment condition can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cable fault diagnosis, and particularly relates to a cable state analysis method and device based on multi-fusion, a terminal device and a storage medium. BACKGROUND

[0002] As an important part of power transmission, the operation state of a cable is directly related to the safety and stability of a power system. With the continuous expansion of the scale of the power system and the increasing complexity of the cable operation environment, the traditional single-parameter cable state evaluation method has been difficult to meet the actual demand. The multi-parameter fusion cable state evaluation method can more comprehensively and accurately reflect the real state of the cable by comprehensively considering various operation parameters of the cable, so as to timely discover potential faults and prevent accidents.

[0003] The current cable state evaluation method still has problems: irrationality of multi-parameter fusion weight determination. In the multi-parameter fusion process, how to determine the weight of a parameter is a key problem. The traditional fixed weight determination method according to expert evaluation is difficult to adapt to the changes of the cable operation condition and environmental condition. SUMMARY

[0004] The present application provides a multi-fusion cable state analysis method, device, terminal device and storage medium, which can solve the problem that the prior art is difficult to adapt to the changes of the cable operation condition and environmental condition.

[0005] An embodiment of the present application provides a multi-fusion cable state analysis method, comprising:

[0006] obtaining cable operation time sequence data and an initial fusion weight vector of a DSFF module; wherein the cable operation time sequence data comprises cable partial discharge time sequence signals, cable dielectric loss time sequence data, cable circulating current time sequence data, cable temperature time sequence data, environmental temperature time sequence data and cable load current time sequence data;

[0007] extracting a plurality of cable operation feature data according to the cable operation time sequence data, and performing normalization processing on the cable operation feature data;

[0008] When at least one cable operation feature data exceeds the corresponding threshold range, according to the normalized cable operation feature data and the initial fusion weight vector, the correlation coefficient of each cable operation feature data and the overall cable operation state is calculated; according to the correlation coefficient, a correlation degree vector is constructed; according to the correlation coefficient, the attention weight of each cable operation state feature data is calculated; according to the attention weight, the corresponding attention weight coefficient is given to each element in the correlation degree vector to obtain an importance weighted vector; according to the importance weighted vector, the initial fusion weight vector and the preset learning rate, the fusion weight vector of the DSFF module is updated;

[0009] The cable operation feature data is input into the DSFF module, so that the DSFF module fuses the cable operation feature data according to the updated fusion weight vector to obtain fusion feature data;

[0010] The fusion feature is input into the trained cable state analysis model, so that the cable state analysis model predicts the cable state according to the fusion to obtain a cable state prediction result.

[0011] Further, the setting method of the initial fusion weight vector of the DSFF module comprises:

[0012] Obtain a plurality of historical cable operation data of the cable operation under different interference environments and at different time points; wherein the historical cable operation data includes: cable partial discharge signal, cable dielectric loss data, cable circulating current data, cable temperature data, environmental temperature data and cable load current data;

[0013] According to the cable operation time series data, a plurality of historical cable operation feature data are extracted, and the historical cable operation feature data are normalized;

[0014] Obtain the relative importance score between each pair of historical cable operation feature data; wherein the relative importance score is obtained by expert scoring;

[0015] According to the relative importance score, a judgment matrix is constructed;

[0016] The characteristic vector corresponding to the maximum eigenvalue of the judgment matrix is calculated by the combined product method, and the characteristic vector is normalized to obtain the first initial weight of each historical cable operation feature data;

[0017] According to the normalized historical cable operation feature data, the entropy value of each historical cable operation feature data is calculated;

[0018] According to the entropy values of all cable operation parameters, the second initial weight of each historical cable operation feature data is calculated;

[0019] According to the first initial weight and the second initial weight, an initial fusion weight of each historical cable operation characteristic data is calculated;

[0020] According to the initial fusion weight of all historical cable operation characteristic data, the initial fusion weight vector is generated;

[0021] Wherein, the entropy value calculation formula of the cable operation parameter is:

[0022]

[0023] In the formula, e j represents the entropy value of the jth historical cable operation characteristic data; m represents the number of historical cable operation characteristic data; z ij represents the jth historical cable operation characteristic data corresponding to the ith historical cable operation data;

[0024] The calculation formula of the second initial weight is:

[0025]

[0026] In the formula, w 2,j represents the second initial weight of the jth historical cable operation characteristic data; n represents the number of historical cable operation characteristic data;

[0027] The calculation formula of the initial fusion weight is:

[0028] w j = β·w 1,j +(1-β)w 2,j ;

[0029] In the formula, w j represents the initial fusion weight of the jth historical cable operation characteristic data; β represents the adjustment coefficient.

[0030] Further, the cable operation characteristic data includes: local discharge signal time domain characteristic mean value, local discharge signal time domain characteristic variance, local discharge signal time domain characteristic peak value, local discharge signal time domain characteristic kurtosis, local discharge signal time domain characteristic skewness, local discharge signal frequency domain characteristic spectrum energy, local discharge signal frequency domain characteristic spectrum entropy, local discharge signal frequency domain characteristic main frequency, local discharge signal time-frequency domain characteristic wavelet transform, local discharge signal time-frequency domain characteristic Hilbert Huang transform, cable dielectric loss factor, cable dielectric loss conversion rate, cable circulating current amplitude, cable circulating current conversion rate, cable temperature mean value, cable temperature variance, cable temperature conversion rate, environmental temperature mean value, environmental temperature variance, environmental temperature conversion rate, cable load current mean value, cable load current variance and cable load current harmonic component.

[0031] Further, the step of calculating the correlation coefficient between each cable operating characteristic data and the overall cable operating status based on the normalized cable operating characteristic data and the initial fusion weight vector includes:

[0032] For each cable operation characteristic data, the modulus of the product of the normalized cable operation characteristic data and the initial fusion weight vector is calculated and used as the correlation coefficient between the cable operation characteristic data and the overall cable operation status.

[0033] The formula for calculating the correlation coefficient is as follows:

[0034] c j =|f j ·W|;

[0035] In the formula, c j f represents the correlation coefficient between the j-th cable operating characteristic data and the overall cable operating status; j represents the j-th cable operating characteristic data after normalization; W represents the initial fusion weight vector.

[0036] Furthermore, the formula for calculating the attention weight is:

[0037]

[0038] In the formula, α j The attention weight represents the feature data of the j-th cable's operating status; c j The correlation coefficient between the j-th cable operating characteristic data and the overall cable operating status is represented by m; m represents the number of cable operating characteristic data.

[0039] Further, the step of assigning a corresponding attention weight coefficient to each element in the relevance vector based on the attention weight to obtain an importance-weighted vector includes:

[0040] For each element in the correlation vector, the product of the attention weight of the cable operation status feature data corresponding to the element and the correlation coefficient is used as the attention weight coefficient of the element.

[0041] The importance weighted vector is generated based on the attention weight coefficients of all elements in the relevance vector.

[0042] Further, updating the fusion weight vector of the DSFF module based on the importance weight vector, the initial fusion weight vector, and the preset learning rate includes:

[0043] The target fusion weight vector is calculated based on the importance weight vector, the initial fusion weight vector, and the preset learning rate.

[0044] Update the fusion weight vector of the DSFF module according to the target fusion weight vector;

[0045] The formula for calculating the target fusion weight vector is as follows:

[0046] W′=W+η·(BW);

[0047] In the formula, W′ represents the target fusion weight vector; W represents the initial fusion weight vector; η represents the learning rate; and B represents the importance weight vector.

[0048] Another embodiment of the present invention provides a cable condition analysis device based on multi-fusion, including: a data acquisition module, a feature extraction module, a fusion weight update module, a feature fusion module, and a cable condition prediction module;

[0049] The data acquisition module is used to acquire cable running timing data and the initial fusion weight vector of the DSFF module; wherein, the cable running timing data includes: cable partial discharge timing signal, cable dielectric loss timing data, cable circulating current timing data, cable temperature timing data, ambient temperature timing data, and cable load current timing data;

[0050] The feature extraction module is used to extract several cable operation feature data based on the cable operation sequence data, and to normalize the cable operation feature data.

[0051] The fusion weight update module is used to calculate the correlation coefficient between each cable operation feature data and the overall cable operation status based on the normalized cable operation feature data and the initial fusion weight vector when at least one cable operation feature data exceeds the corresponding threshold range; construct an association degree vector based on the correlation coefficient; calculate the attention weight of each cable operation status feature data based on the correlation coefficient; assign a corresponding attention weight coefficient to each element in the association degree vector based on the attention weight to obtain an importance weighted vector; and update the fusion weight vector of the DSFF module based on the importance weighted vector, the initial fusion weight vector, and a preset learning rate.

[0052] The feature fusion module is used to input the cable operation feature data into the DSFF module, so that the DSFF module fuses the cable operation feature data according to the updated fusion weight vector to obtain fused feature data;

[0053] The cable condition prediction module is used to input the fused features into the trained cable condition analysis model, so that the cable condition analysis model can predict the cable condition based on the fusion and obtain the cable condition prediction result.

[0054] Another embodiment of the present invention provides a terminal device, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the steps of the cable condition analysis method based on multi-parameter fusion of the present invention.

[0055] Another embodiment of the present invention provides a computer-readable storage medium item, including: a stored computer program, which, when the computer program is running, controls the device where the computer-readable storage medium is located to perform the steps of the cable condition analysis method based on multi-parameter fusion of the present invention.

[0056] The following benefits can be obtained by implementing the present invention:

[0057] This invention acquires cable operation timing data and an initial fusion weight vector for the DSFF module. The cable operation timing data includes: cable partial discharge timing signal, cable dielectric loss timing data, cable circulating current timing data, cable temperature timing data, ambient temperature timing data, and cable load current timing data. Based on the cable operation timing data, several cable operation feature data are extracted and normalized. When at least one cable operation feature data exceeds a corresponding threshold range, the correlation coefficient between each cable operation feature data and the overall cable operation state is calculated based on the normalized cable operation feature data and the initial fusion weight vector. A correlation degree vector is constructed based on the correlation coefficient. The correlation coefficient is used to calculate the attention weight of each cable operating status feature data. Based on the attention weight, each element in the correlation vector is assigned a corresponding attention weight coefficient to obtain an importance weighted vector. The fusion weight vector of the DSFF module is updated based on the importance weighted vector, the initial fusion weight vector, and a preset learning rate. The cable operating feature data is input into the DSFF module, so that the DSFF module fuses the cable operating feature data according to the updated fusion weight vector to obtain fused feature data. The fused features are input into a trained cable state analysis model, so that the cable state analysis model predicts the cable state based on the fusion to obtain a cable state prediction result. The cable operating sequence data acquired in real time by this invention includes data related to cable operating conditions and data related to environmental conditions. Feature data is extracted from the cable operating sequence data, and the dynamic adjustment mechanism of the fusion weight vector is triggered based on the feature data to make the adjusted fusion weight vector more adaptable to the current cable operating conditions and environmental conditions. Attached Figure Description

[0058] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0059] Figure 1 This is a flowchart illustrating a cable condition analysis method based on multi-parameter fusion provided in an embodiment of the present invention;

[0060] Figure 2 This is a schematic diagram of a cable condition analysis device based on multi-parameter fusion provided in an embodiment of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the term "comprising" and any variations thereof in the specification, claims and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0063] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "several" means two or more, unless otherwise explicitly defined.

[0064] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0065] See Figure 1To address the problem that existing technologies are ill-suited to adapting to changes in cable operating conditions and environmental conditions, an embodiment of the present invention provides a cable condition analysis method based on multi-fusion, comprising:

[0066] S1. Obtain cable running timing data and the initial fusion weight vector of the DSFF module; wherein, the cable running timing data includes: cable partial discharge timing signal, cable dielectric loss timing data, cable circulating current timing data, cable temperature timing data, ambient temperature timing data, and cable load current timing data.

[0067] In step S1, multiple sets of cable operation sequence data at different time points are collected in real time using various sensors. Each set includes partial discharge, dielectric loss, circulating current, temperature, and load current. High-voltage coupling capacitors are installed at key locations such as cable joints and terminals to monitor partial discharge signals in real time. A high-voltage dielectric loss tester is used to measure the cable's dielectric loss factor, and data is collected periodically. A current transformer is installed on the cable's grounding wire to measure the cable's circulating current. Fiber optic temperature sensors are installed along the cable's length and in the surrounding environment to monitor the cable's temperature and ambient temperature. Current transformers are used to measure the cable's load current and are installed at the cable's inlet and outlet ends. The collected data undergoes cleaning, normalization, and time alignment to improve data quality and usability. During data cleaning, outliers and noise are removed, and missing values ​​are processed. Next, data normalization is performed to normalize data of different dimensions and magnitudes to the same range for subsequent processing and analysis. Finally, time alignment is performed to ensure that data collected by different sensors are aligned in time for fusion analysis.

[0068] In a preferred embodiment, the method for setting the initial fusion weight vector of the DSFF module includes:

[0069] Acquire several historical cable operation data at different times under different interference environments; wherein, the historical cable operation data includes: cable partial discharge signal, cable dielectric loss data, cable circulating current data, cable temperature data, ambient temperature data, and cable load current data;

[0070] Based on the cable operation sequence data, several historical cable operation feature data are extracted, and the historical cable operation feature data are normalized.

[0071] Obtain the relative importance score between pairs of historical cable operation characteristic data; wherein, the relative importance score is obtained by expert scoring;

[0072] Based on the relative importance scores, construct a judgment matrix;

[0073] The eigenvector corresponding to the largest eigenvalue of the judgment matrix is ​​calculated by the sum product method, and the eigenvector is normalized to obtain the first initial weight of each historical cable operation characteristic data.

[0074] Based on the normalized historical cable operation characteristic data, calculate the entropy value of each historical cable operation characteristic data.

[0075] Calculate the second initial weight for each historical cable operating characteristic data based on the entropy values ​​of all cable operating parameters;

[0076] Based on the first initial weight and the second initial weight, calculate the initial fusion weight of each historical cable operation characteristic data;

[0077] The initial fusion weight vector is generated based on the initial fusion weights of all historical cable operation characteristic data;

[0078] The entropy value of the cable operating parameters is calculated using the following formula:

[0079]

[0080] In the formula, e j The z represents the entropy value of the j-th historical cable operation characteristic data; m represents the number of historical cable operation characteristic data; z ij This represents the historical cable operation characteristic data corresponding to the i-th historical cable operation data;

[0081] The formula for calculating the second initial weight is:

[0082]

[0083] In the formula, w 2,j The second initial weight represents the j-th historical cable operation characteristic data; n represents the number of historical cable operation characteristic data.

[0084] The formula for calculating the initial fusion weight is:

[0085] w j =β·w 1,j +(1-β)w 2,j ;

[0086] In the formula, w j β represents the initial fusion weight of the j-th historical cable operation characteristic data; β represents the adjustment coefficient.

[0087] In this embodiment, a laboratory cable operation status data acquisition platform is constructed to comprehensively collect various operating parameters of the cable while simulating several industrial field interference environments as much as possible.

[0088] The collected data is cleaned, normalized, and time-aligned to improve its quality and usability.

[0089] A hierarchical model containing various feature data was constructed using the Analytic Hierarchy Process (AHP). The importance of each feature data point was compared pairwise by expert evaluation. Experts scored the relative importance of each feature data point based on their experience and expertise, and a judgment matrix was constructed using a 1-9 scale to compare the importance of two feature data points: 1 for "equally important," 3 for "slightly important," 5 for "significantly important," 7 for "strongly important," and 9 for "extremely important." Next, the eigenvector corresponding to the largest eigenvalue of the matrix was calculated using the summation method, and normalized to obtain the AHP weight vector, which was denoted as the first initial weight vector for each parameter. Finally, the consistency index and consistency ratio of the judgment matrix were calculated to verify its consistency.

[0090] The entropy value of each feature data is calculated using the entropy weight method. The magnitude of the entropy value reflects the uncertainty of the parameter; the smaller the entropy value, the greater the information content of the parameter and the greater its contribution to the cable condition assessment. Then, a second initial weight is calculated based on the entropy value of each feature data.

[0091] The initial weights obtained from the analytic hierarchy process (AHP) and the entropy weight method are combined to obtain the initial fused weights. The adjustment parameters during fusion can be adjusted according to actual needs. This method, which combines subjective and objective approaches, can provide a more comprehensive and accurate initial fused weights.

[0092] S2. Based on the cable operation sequence data, extract several cable operation characteristic data and normalize the cable operation characteristic data.

[0093] In step S2, features are extracted based on the different types of parameter data that have undergone data cleaning, normalization, and time alignment in step S1, using appropriate methods.

[0094] In a preferred embodiment, the cable operating characteristic data includes: the mean value of the partial discharge signal in the time domain, the variance of the partial discharge signal in the time domain, the peak value of the partial discharge signal in the time domain, the kurtosis of the partial discharge signal in the time domain, the skewness of the partial discharge signal in the time domain, the spectral energy of the partial discharge signal in the frequency domain, the spectral entropy of the partial discharge signal in the frequency domain, the dominant frequency of the partial discharge signal in the frequency domain, the wavelet transform of the partial discharge signal in the time-frequency domain, the Hilbert-Huang transform of the partial discharge signal in the time-frequency domain, the cable dielectric loss factor, the cable dielectric loss transformation rate, the cable circulating current amplitude, the cable circulating current transformation rate, the mean value of the cable temperature, the cable temperature variance, the cable temperature transformation rate, the mean value of the ambient temperature, the ambient temperature variance, the ambient temperature transformation rate, the mean value of the cable load current, the variance of the cable load current, and the harmonic components of the cable load current.

[0095] In this embodiment, partial discharge signals are characterized from time-domain, frequency-domain, and time-frequency perspectives. Time-domain features include mean, variance, peak value, kurtosis, and skewness, reflecting the statistical characteristics of the partial discharge signal. Frequency-domain features include spectral energy, spectral entropy, and dominant frequency, capturing the signal's frequency distribution. Time-frequency features include multi-scale features such as wavelet transform and Hilbert transform, analyzing the signal's time-frequency variations. Dielectric loss features are extracted, primarily including the dielectric loss factor and its transformation rate, to reflect insulation aging trends. Circulating current features are extracted, primarily including the magnitude of the circulating current and its transformation rate, to detect dynamic changes in the circulating current. Temperature features are extracted, primarily including the mean, variance, and transformation rate of temperature, to comprehensively reflect temperature fluctuations and their impact on cable condition. Load current features are extracted, primarily including the mean, variance, and harmonic components, to capture load current fluctuations and distortions, thereby assessing the cable's load condition and potential overload risk.

[0096] S3. When at least one cable operation feature data exceeds the corresponding threshold range, calculate the correlation coefficient between each cable operation feature data and the overall cable operation status based on the normalized cable operation feature data and the initial fusion weight vector; construct a correlation degree vector based on the correlation coefficient; calculate the attention weight of each cable operation status feature data based on the correlation coefficient; assign a corresponding attention weight coefficient to each element in the correlation degree vector based on the attention weight to obtain an importance weighted vector; update the fusion weight vector of the DSFF module based on the importance weighted vector, the initial fusion weight vector, and the preset learning rate.

[0097] In step S3, a dynamic adjustment mechanism is introduced to adapt to changes in cable operating status and environmental conditions. This involves real-time monitoring of the cable's operating status and environmental conditions, setting thresholds for each characteristic data point, and determining whether any characteristic data exceeds the threshold (e.g., temperature exceeding the average by 5°C, circulating current amplitude increasing by more than 20%). If any of these thresholds are exceeded, the dynamic adjustment mechanism is triggered, automatically adjusting the weight of each characteristic data point based on its importance.

[0098] In a preferred embodiment, calculating the correlation coefficient between each cable operating characteristic data point and the overall cable operating status based on the normalized cable operating characteristic data and the initial fusion weight vector includes:

[0099] For each cable operation characteristic data, the modulus of the product of the normalized cable operation characteristic data and the initial fusion weight vector is calculated and used as the correlation coefficient between the cable operation characteristic data and the overall cable operation status.

[0100] The formula for calculating the correlation coefficient is as follows:

[0101] c j =|f j ·W|;

[0102] In the formula, c j f represents the correlation coefficient between the j-th cable operating characteristic data and the overall cable operating status; j represents the j-th cable operating characteristic data after normalization; W represents the initial fusion weight vector.

[0103] In this embodiment, the correlation coefficient between feature data and the overall cable operating status is calculated using an attention mechanism.

[0104] In a preferred embodiment, the formula for calculating the attention weight is:

[0105]

[0106] In the formula, α j The attention weight represents the feature data of the j-th cable's operating status; c j The correlation coefficient between the j-th cable operating characteristic data and the overall cable operating status is represented by m; m represents the number of cable operating characteristic data.

[0107] In this embodiment, the softmax function is used to convert the correlation coefficient into attention weights, ensuring that the sum of the weights is 1, thus highlighting the influence of highly correlated features.

[0108] In a preferred embodiment, the step of assigning a corresponding attention weight coefficient to each element in the relevance vector according to the attention weight to obtain an importance-weighted vector includes:

[0109] For each element in the correlation vector, the product of the attention weight of the cable operation status feature data corresponding to the element and the correlation coefficient is used as the attention weight coefficient of the element.

[0110] The importance weighted vector is generated based on the attention weight coefficients of all elements in the relevance vector.

[0111] It should be noted that the importance-weighted vector is a feature importance-weighted vector based on attention.

[0112] In a preferred embodiment, updating the fusion weight vector of the DSFF module based on the importance weight vector, the initial fusion weight vector, and a preset learning rate includes:

[0113] The target fusion weight vector is calculated based on the importance weight vector, the initial fusion weight vector, and the preset learning rate.

[0114] Update the fusion weight vector of the DSFF module according to the target fusion weight vector;

[0115] The formula for calculating the target fusion weight vector is as follows:

[0116] W′=W+η·(BW);

[0117] In the formula, W′ represents the target fusion weight vector; W represents the initial fusion weight vector; η represents the learning rate; and B represents the importance weight vector.

[0118] In this embodiment, the fusion weights are updated based on an importance-weighted vector. Less important connections are gradually weakened or even pruned, while important connections are strengthened. The target fusion weight vector is not directly used as input to the cable condition analysis model, but rather indirectly affects the input and output of the model by adjusting the fusion weights of the feature matrix.

[0119] S4. Input the cable operation characteristic data into the DSFF module so that the DSFF module fuses the cable operation characteristic data according to the updated fusion weight vector to obtain fused characteristic data.

[0120] S5. Input the fused features into the trained cable condition analysis model so that the cable condition analysis model can predict the cable condition based on the fusion and obtain the cable condition prediction result.

[0121] It should be noted that the cable condition analysis model is the BiTCN-Transformer model.

[0122] In step S5, the fused feature data is input into the cable condition analysis model to predict the cable condition. Features at different time scales are extracted through different convolutional layers of the BiTCN module, and different spatial correlation features are generated through the multi-head attention mechanism of the Transformer module, forming a multi-scale feature map set. Then, the set of multi-scale feature maps is fused to construct a comprehensive feature map. Finally, the channel dimension is fused through the convolutional layer feature maps to generate a comprehensive feature map with a unified dimension, capturing the spatiotemporal correlation features of the cable condition.

[0123] In a preferred embodiment, the training process of the cable condition analysis model includes:

[0124] Obtain several historical cable operation data samples;

[0125] Based on the historical cable operation data samples, several historical cable operation feature data samples are extracted.

[0126] The labels are created by experts based on the actual condition of the cables and include historical cable operation characteristic data samples; the types of labels include "normal", "warning", and "fault".

[0127] The parameters of the BiTCN-Transformer model are updated iteratively using the cross-entropy loss function and the Adam optimizer.

[0128] It should be noted that the parameters updated in the model include the convolutional kernel weights of BiTCN and the multi-head attention matrix weights of Transformer. The model output is a probability distribution P = [p1, p2, p3] of the cable state, corresponding to the three states of "normal", "warning", and "fault" respectively. The state with the highest probability is taken as the cable state prediction result.

[0129] like Figure 2 As shown, based on the above method embodiments, corresponding apparatus embodiments are provided;

[0130] An embodiment of the present invention provides a cable condition analysis device based on multi-fusion, including: a data acquisition module, a feature extraction module, a fusion weight update module, a feature fusion module, and a cable condition prediction module;

[0131] The data acquisition module is used to acquire cable running timing data and the initial fusion weight vector of the DSFF module; wherein, the cable running timing data includes: cable partial discharge timing signal, cable dielectric loss timing data, cable circulating current timing data, cable temperature timing data, ambient temperature timing data, and cable load current timing data;

[0132] The feature extraction module is used to extract several cable operation feature data based on the cable operation sequence data, and to normalize the cable operation feature data.

[0133] The fusion weight update module is used to calculate the correlation coefficient between each cable operation feature data and the overall cable operation status based on the normalized cable operation feature data and the initial fusion weight vector when at least one cable operation feature data exceeds the corresponding threshold range; construct an association degree vector based on the correlation coefficient; calculate the attention weight of each cable operation status feature data based on the correlation coefficient; assign a corresponding attention weight coefficient to each element in the association degree vector based on the attention weight to obtain an importance weighted vector; and update the fusion weight vector of the DSFF module based on the importance weighted vector, the initial fusion weight vector, and a preset learning rate.

[0134] The feature fusion module is used to input the cable operation feature data into the DSFF module, so that the DSFF module fuses the cable operation feature data according to the updated fusion weight vector to obtain fused feature data;

[0135] The cable condition prediction module is used to input the fused features into the trained cable condition analysis model, so that the cable condition analysis model can predict the cable condition based on the fusion and obtain the cable condition prediction result.

[0136] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can implement the cable condition analysis method based on multi-parameter fusion provided by any of the above-described method embodiments of the present invention.

[0137] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0138] Based on the above-described method embodiments, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the cable status analysis method based on multi-parameter fusion of any embodiment of the present invention.

[0139] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0140] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0141] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0142] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the cable condition analysis method based on multi-parameter fusion as described in any of the above-described method embodiments of the present invention.

[0143] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0144] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A cable condition analysis method based on multi-fusion, characterized in that, include: Obtain cable runtime timing data and the initial fusion weight vector of the DSFF module; wherein, the cable runtime timing data includes: cable partial discharge timing signal, cable dielectric loss timing data, cable circulating current timing data, cable temperature timing data, ambient temperature timing data, and cable load current timing data; Based on the cable operation sequence data, several cable operation characteristic data are extracted, and the cable operation characteristic data are normalized. When at least one cable operation feature data exceeds the corresponding threshold range, the correlation coefficient between each cable operation feature data and the overall cable operation status is calculated based on the normalized cable operation feature data and the initial fusion weight vector; an association degree vector is constructed based on the correlation coefficient; the attention weight of each cable operation status feature data is calculated based on the correlation coefficient; each element in the association degree vector is assigned a corresponding attention weight coefficient based on the attention weight to obtain an importance weighted vector; the fusion weight vector of the DSFF module is updated based on the importance weighted vector, the initial fusion weight vector, and the preset learning rate. The cable operation characteristic data is input into the DSFF module, so that the DSFF module fuses the cable operation characteristic data according to the updated fusion weight vector to obtain fused characteristic data; The fused features are input into the trained cable condition analysis model so that the cable condition analysis model can predict the cable condition based on the fusion, and obtain the cable condition prediction result.

2. The cable condition analysis method based on multi-parameter fusion as described in claim 1, characterized in that, The method for setting the initial fusion weight vector of the DSFF module includes: Acquire several historical cable operation data at different times under different interference environments; wherein, the historical cable operation data includes: cable partial discharge signal, cable dielectric loss data, cable circulating current data, cable temperature data, ambient temperature data, and cable load current data; Based on the cable operation sequence data, several historical cable operation feature data are extracted, and the historical cable operation feature data are normalized. Obtain the relative importance score between pairs of historical cable operation characteristic data; wherein, the relative importance score is obtained by expert scoring; Based on the relative importance scores, construct a judgment matrix; The eigenvector corresponding to the largest eigenvalue of the judgment matrix is ​​calculated by the sum product method, and the eigenvector is normalized to obtain the first initial weight of each historical cable operation characteristic data. Based on the normalized historical cable operation characteristic data, calculate the entropy value of each historical cable operation characteristic data. Calculate the second initial weight for each historical cable operating characteristic data based on the entropy values ​​of all cable operating parameters; Based on the first initial weight and the second initial weight, calculate the initial fusion weight of each historical cable operation characteristic data; The initial fusion weight vector is generated based on the initial fusion weights of all historical cable operation characteristic data; The entropy value of the cable operating parameters is calculated using the following formula: In the formula, e j The z represents the entropy value of the j-th historical cable operation characteristic data; m represents the number of historical cable operation characteristic data; z ij This represents the historical cable operation characteristic data corresponding to the i-th historical cable operation data; The formula for calculating the second initial weight is: In the formula, w 2,j The second initial weight represents the j-th historical cable operation characteristic data; n represents the number of historical cable operation characteristic data. The formula for calculating the initial fusion weight is: w j =β·w 1,j +(1-β)w 2,j ; In the formula, w j β represents the initial fusion weight of the j-th historical cable operation characteristic data; β represents the adjustment coefficient.

3. The cable condition analysis method based on multi-parameter fusion as described in claim 2, characterized in that, The cable operating characteristic data includes: mean value of partial discharge signal in the time domain, variance value of partial discharge signal in the time domain, peak value value of partial discharge signal in the time domain, kurtosis value of partial discharge signal in the time domain, skewness value of partial discharge signal in the time domain, spectral energy value of partial discharge signal in the frequency domain, spectral entropy value of partial discharge signal in the frequency domain, dominant frequency value of partial discharge signal in the frequency domain, wavelet transform value of partial discharge signal in the time-frequency domain, Hilbert-Huang transform value of partial discharge signal in the time-frequency domain, cable dielectric loss factor, cable dielectric loss transformation rate, cable circulating current amplitude, cable circulating current transformation rate, mean value of cable temperature, cable temperature variance, cable temperature transformation rate, mean value of ambient temperature, ambient temperature variance, ambient temperature transformation rate, mean value of cable load current, variance of cable load current, and harmonic components of cable load current.

4. The cable condition analysis method based on multi-parameter fusion as described in claim 1, characterized in that, The step of calculating the correlation coefficient between each cable operating characteristic data and the overall cable operating status based on the normalized cable operating characteristic data and the initial fusion weight vector includes: For each cable operation characteristic data, the modulus of the product of the normalized cable operation characteristic data and the initial fusion weight vector is calculated and used as the correlation coefficient between the cable operation characteristic data and the overall cable operation status. The formula for calculating the correlation coefficient is as follows: c j =|f j ·W|; In the formula, c j f represents the correlation coefficient between the j-th cable operating characteristic data and the overall cable operating status; j represents the j-th cable operating characteristic data after normalization; W represents the initial fusion weight vector.

5. The cable condition analysis method based on multi-parameter fusion as described in claim 1, characterized in that, The formula for calculating the attention weight is: In the formula, α j This represents the attention weight of the j-th cable's operating status feature data; c j The correlation coefficient between the j-th cable operating characteristic data and the overall cable operating status is represented by m; m represents the number of cable operating characteristic data.

6. The cable condition analysis method based on multi-parameter fusion as described in claim 1, characterized in that, The step of assigning a corresponding attention weight coefficient to each element in the relevance vector according to the attention weight to obtain an importance-weighted vector includes: For each element in the correlation vector, the product of the attention weight of the cable operation status feature data corresponding to the element and the correlation coefficient is used as the attention weight coefficient of the element. The importance weighted vector is generated based on the attention weight coefficients of all elements in the relevance vector.

7. The cable condition analysis method based on multi-parameter fusion as described in claim 1, characterized in that, The step of updating the fusion weight vector of the DSFF module based on the importance weight vector, the initial fusion weight vector, and the preset learning rate includes: The target fusion weight vector is calculated based on the importance weight vector, the initial fusion weight vector, and the preset learning rate. Update the fusion weight vector of the DSFF module according to the target fusion weight vector; The formula for calculating the target fusion weight vector is as follows: W′=W+η·(BW); In the formula, W′ represents the target fusion weight vector; W represents the initial fusion weight vector; η represents the learning rate; and B represents the importance weight vector.

8. A cable condition analysis device based on multi-fusion, characterized in that, include: The module includes a data acquisition module, a feature extraction module, a fusion weight update module, a feature fusion module, and a cable status prediction module. The data acquisition module is used to acquire cable running timing data and the initial fusion weight vector of the DSFF module; wherein, the cable running timing data includes: cable partial discharge timing signal, cable dielectric loss timing data, cable circulating current timing data, cable temperature timing data, ambient temperature timing data, and cable load current timing data; The feature extraction module is used to extract several cable operation feature data based on the cable operation sequence data, and to normalize the cable operation feature data. The fusion weight update module is used to calculate the correlation coefficient between each cable operation feature data and the overall cable operation status based on the normalized cable operation feature data and the initial fusion weight vector when at least one cable operation feature data exceeds the corresponding threshold range; construct an association degree vector based on the correlation coefficient; calculate the attention weight of each cable operation status feature data based on the correlation coefficient; assign a corresponding attention weight coefficient to each element in the association degree vector based on the attention weight to obtain an importance weighted vector; and update the fusion weight vector of the DSFF module based on the importance weighted vector, the initial fusion weight vector, and a preset learning rate. The feature fusion module is used to input the cable operation feature data into the DSFF module, so that the DSFF module fuses the cable operation feature data according to the updated fusion weight vector to obtain fused feature data; The cable condition prediction module is used to input the fused features into the trained cable condition analysis model, so that the cable condition analysis model can predict the cable condition based on the fusion and obtain the cable condition prediction result.

9. A terminal device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the cable condition analysis method based on multi-parameter fusion as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, include: A stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the cable condition analysis method based on multi-parameter fusion as described in any one of claims 1-7.

Images