Cable monitoring method based on data fusion
Through data fusion technology, the use of current harmonics and strain data, combined with convolutional neural network, the accurate monitoring of the deterioration of cables is achieved, and the problem of single and poor universality of cable monitoring parameters in the existing technology is solved, and the more Supports reliable cable operation and maintenance.
Patent Information
- Application Number
- CN202411147883.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-26
AI Technical Summary
Existing cable monitoring methods cannot effectively monitor the overall insulation status and deterioration of the cable, and there are problems of single and insufficient parameters, making it difficult to accurately evaluate the damage to the cable.
The cable monitoring method based on data fusion is adopted, and the current harmonic data and strain data of the cable are obtained through preset monitoring nodes, a database and an impact factor library are established, and the data is preprocessed and classified using a convolutional neural network to output the cable monitoring results.
It realizes non-invasive cable monitoring, can accurately evaluate the deterioration of cables, overcomes the problems of single parameters and poor universality, and provides more reliable cable operation and maintenance guidance.
Smart Images

Figure CN119024227A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of monitoring and control, and specifically discloses a cable monitoring method based on data fusion. Background Art
[0002] In the electric power industry, cable defects are of various types and operating conditions are complex. It is of great significance to improve the reliability of distribution network operation to discover and identify cable defects and adverse environments around cables as early as possible, eliminate hidden dangers, repair or replace them in time.
[0003] At present, there are some monitoring methods for the cable environment, such as installing a monitoring system in the cable tunnel to monitor the tunnel cables. However, the monitoring has certain disadvantages. It can monitor theft, but it cannot monitor the cable status affected by environmental factors, and lacks reliable monitoring of the normal operation status of the cable.
[0004] Cable monitoring: The mainstream cable status detection methods at home and abroad, such as the partial discharge method, are the most commonly used cable status detection methods, but their limitations are that they cannot detect circuits that have not damaged the semi-conductive layer and the main insulation layer, and it is difficult to evaluate the overall insulation status of the cable. The dielectric loss tangent method can be used for on-site live detection, but the insulation parameters obtained are relatively simple and cannot accurately evaluate the degradation of the cable. The above mainstream detection methods have room for improvement. Summary of the invention
[0005] The first objective of the embodiment of the present invention is to provide a cable monitoring method based on data fusion, aiming to solve the problems mentioned in the background technology.
[0006] The embodiment of the present invention is implemented as follows: a cable monitoring method based on data fusion, the method comprising:
[0007] Step S1, obtaining the current harmonic data of the cable through a preset monitoring node for monitoring the cable, and establishing a current harmonic database;
[0008] Step S2, obtaining the strain data of the cable at the monitoring node and the displacement change data of the monitoring node, and establishing a node influence factor library;
[0009] Step S3, aligning the current harmonic database and the node influence factor database according to the acquisition time and the monitoring node position, and preprocessing the aligned fusion data set;
[0010] Step S4: classify the degradation of the fused data set through the trained cable convolutional neural network, and output the cable monitoring results according to the classification.
[0011] The second object of the present invention is to provide a cable monitoring system based on data fusion, which is used in the cable monitoring method based on data fusion, and the system comprises:
[0012] The first database module is used to obtain the current harmonic data of the cable through a preset monitoring node for monitoring the cable, and establish a current harmonic database;
[0013] The second database module is used to obtain the strain data of the cable at the monitoring node and the displacement change data of the monitoring node, and establish a node influence factor library;
[0014] The data fusion module is used to align the current harmonic database and the node influence factor database according to the acquisition time and the monitoring node location, and pre-process the aligned fusion data set;
[0015] The cable classification module is used to classify the degradation of the fused data set through the trained cable convolutional neural network and output the cable monitoring results according to the classification.
[0016] The third object of the present invention is to provide a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, steps S1 to S4 of the cable monitoring method based on data fusion are implemented.
[0017] The fourth object of the present invention is to provide a computer-readable storage medium storing computer instructions or computer programs, which, when executed by one or more processors, cause the processors to execute steps S1 to S4 of the cable monitoring method based on data fusion.
[0018] A cable monitoring method based on data fusion provided by an embodiment of the present invention can obtain relevant data for monitoring in a non-invasive manner compared to the existing cable monitoring method; it can overcome the defects of being unable to conduct on-site live detection or the relatively single monitoring parameters obtained, and being unable to accurately assess the damage and deterioration of the cable; selecting appropriate monitoring nodes according to the actual laying of the cable, for example: selecting the location or vicinity of the cable connection joint or connector; the current harmonic data of the cable can be monitored by instruments such as a harmonic analyzer, the strain data of the cable can be obtained by a tension sensor, and the displacement change data of the monitoring node can be obtained by sensors such as a displacement sensor or a gyroscope; after obtaining relevant data through each sensor, the data can be normalized first to facilitate data fusion; thereafter, feature extraction is performed on the fused data set, and the cables are classified according to the extracted features, and the classification results are saved, so as to quantitatively display the cable monitoring and guide the operation and maintenance of the cable, and can also solve the problems of single, insufficient and poor universality of parameters in the prior art when detecting cables. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 A flow chart of a cable monitoring method based on data fusion provided by an embodiment of the present invention;
[0020] Figure 2 A flow chart of establishing a current harmonic database in one embodiment;
[0021] Figure 3 A flowchart for generating a maintenance request in one embodiment;
[0022] Figure 4 A sub-flow chart for generating a maintenance request in one embodiment;
[0023] Figure 5 A flowchart of classifying cable degradation in one embodiment;
[0024] Figure 6 A sub-flow chart for performing deep feature extraction on current harmonic data in one embodiment;
[0025] Figure 7 A flowchart of training a cable convolutional neural network in one embodiment;
[0026] Figure 8 A sub-flow chart for training a cable convolutional neural network in one embodiment;
[0027] Fig. 9 A flowchart of periodically adjusting a preset position of a monitoring node in one embodiment;
[0028] Fig.10A structural block diagram of a cable monitoring system based on data fusion provided by an embodiment of the present invention;
[0029] Fig.11 FIG. 4 is a block diagram of the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0030] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0031] Figure 1 A flow chart of a cable monitoring method based on data fusion provided by an embodiment of the present invention may specifically include steps S1 to S4;
[0032] Step S1, obtaining the current harmonic data of the cable through a preset monitoring node for monitoring the cable, and establishing a current harmonic database;
[0033] In this step, the monitoring nodes are pre-set. Generally, they can be conveniently selected according to the actual cable laying site, for example: a displacement sensor is set at the selected position in the cable's heat dissipation well, maintenance well, etc., or at or near the cable's connection joint or connector to monitor the displacement changes of the monitoring node or the cable connection; a tension sensor is set to monitor the tension changes of the cable; the specific implementation can refer to relevant standards, for example: GB 50217-2018.
[0034] Step S2, obtaining the strain data of the cable at the monitoring node and the displacement change data of the monitoring node, and establishing a node influence factor library;
[0035] Among them, the deformation of the cable can be inferred by monitoring the displacement changes of the monitoring nodes; and the displacement sensors can be directly set at the relevant positions of the cable to directly obtain the displacement changes of the cable and then infer the deformation of the cable.
[0036] Step S3, aligning the current harmonic database and the node influence factor database according to the acquisition time and the monitoring node position, and preprocessing the aligned fusion data set;
[0037] First, the current harmonic data in the current harmonic database, the strain data in the node influence factor library, and the corresponding displacement change data are aligned for the first time according to the acquisition time; specifically, the acquisition time of each data in the current harmonic database and the node influence factor library is obtained, and the data in the current harmonic database and the node influence factor library are aligned using the acquisition time as the alignment standard;
[0038] Secondly, the current harmonic database and node influence factor library after the first alignment are aligned for the second time according to the monitoring node positions to establish a fused data set.
[0039] Among them, preprocessing the fused data set generally includes data screening, amplification or cleaning, and normalization;
[0040] Step S4, performing degradation classification on the fused data set through the trained cable convolutional neural network, and outputting the cable monitoring results according to the classification;
[0041] In one example, the node influencing factors of the fused data set, namely the strain data and displacement change data, are first used to optimize the parameters of the trained cable convolutional neural network in order to obtain a better cable convolutional neural network; then, the cable convolutional neural network is used to extract features of the current harmonic data in the fused data set and classify cable degradation.
[0042] In step S1, before the harmonic data of the cable current is acquired, the signal is first acquired through a harmonic acquisition device. The main component of the harmonic acquisition device is a voltage and current transformer. The data acquired by the voltage and current transformer is input into a microprocessor for processing after signal conversion and signal conditioning, so as to obtain the current harmonic data of the cable for easy acquisition;
[0043] In practice, it is found that the aging degree of the cable affects the harmonic signal of the cable current. Therefore, when establishing the current harmonic database, some algorithms can be used to filter and screen the signal noise;
[0044] In this embodiment, the degradation of the cable at least includes the outer sheath being damaged, the shielding layer being damaged, and the bending degree being large; thus, in this embodiment, the degradation classification of the cable mainly involves the identification and classification of the outer sheath being damaged, the shielding layer being damaged, and the bending degree being large;
[0045] Among them, the harmonic current content rate is different in the higher harmonics corresponding to the damage of the outer sheath, damage of the shielding layer and the bending degree of the cable. In this way, the extraction features of the damage of the outer sheath, damage of the shielding layer and the bending degree of the cable can be obtained according to the calculation of the harmonic current content rate of each higher harmonic. Conversely, according to the extracted features, it can be judged whether the cable has the damage of the outer sheath, damage of the shielding layer or the bending degree. Of course, the cable may only have one or part of the degradation phenomenon. After classifying the relevant degradation conditions, targeted maintenance can be carried out.
[0046] like Figure 2 As shown, in one embodiment, step S1 includes:
[0047] Step S11, after obtaining the current harmonic data of the cable, determine the fundamental wave of the current harmonic of the cable;
[0048] Step S12, after shifting the basis function of the fundamental wave according to the wavelet transform formula, perform inner product with the signal to be analyzed at different scales to obtain the distribution of higher harmonics;
[0049] Step S13, establishing a current harmonic database according to the distribution of each high-order harmonic;
[0050] Among them, the wavelet transform formula is a direct application of the existing technology and will not be described in detail here.
[0051] In one embodiment, in step S2, strain data of the cable at the monitoring node and displacement change data of the monitoring node are obtained based on a distributed communication network.
[0052] For example: an edge IoT communication module is set up at each monitoring node, and a sensor is connected through a data acquisition card to obtain the strain data of the cable at the monitoring node and the displacement change data of the monitoring node; the data acquisition card can be a mobile Internet communication protocol, an IoT card communication protocol, etc., and the edge IoT communication modules are merged through a multiplexed connector and finally connected to the cloud server to realize the construction of a distributed network.
[0053] like Figure 3 , Figure 4 As shown, in an example of an embodiment, the method further includes: step S210, generating a maintenance request, specifically including:
[0054] Step S211, gridding the distributed communication network and monitoring nodes;
[0055] Among them, the scale of gridding is fixed or variable. The role of gridding is to establish the distributed communication network and monitoring nodes into a map interface, which can be a public map currently available on the market, so as to facilitate the convenient construction, sharing and visual management of monitoring nodes.
[0056] Step S212: judging whether the monitoring node is available according to the communication status of each communication module in the distributed communication network;
[0057] In this way, for a communication module that cannot obtain a communication connection response, it can be considered that the monitoring node where the communication module is located has failed, which can assist in quickly locating the position of the suspicious monitoring node; the suspicious monitoring node can be marked, and specific cable classification can be performed based on the cable harmonic data of the marked cable when calling the cable, with higher accuracy.
[0058] Step S213: When the monitoring node is unavailable, a maintenance request is fed back; the maintenance request includes the device ID and setting address of the communication module corresponding to the monitoring node; the position of the suspicious monitoring node can be quickly located according to the device ID and setting address of the corresponding communication module, so as to provide support for further accurate monitoring and subsequent repair and maintenance;
[0059] When the monitoring node is available, no action is taken.
[0060] In one embodiment, in step S211, after the distributed communication network and monitoring nodes are gridded,
[0061] According to the gridded monitoring nodes, a weight setting directory is established and associated with the cable fault database;
[0062] Based on the weight setting directory, the cable fault classification data in the cable fault database is called to determine the strain weight of the cable and the displacement weight of the monitoring node:
[0063] After assigning the strain weight and displacement weight, the monitoring nodes are iterated to form a node change field;
[0064] The optimal values of strain weight and displacement weight are selected according to the node change field, and the node influence factor library is updated.
[0065] In this embodiment, the cable fault database includes almost all types of conventional cables, such as the above-mentioned outer sheath damage, shielding layer damage or large bends, etc. The cable fault database is generally commonly used by cable manufacturers. Of course, it can also be manually established by senior experts in the power industry, which will not be described in detail here;
[0066] In an example, for example, the node influencing factors include strain and displacement. According to the cable fault category, the strain influence is numerically quantified, and the strain quantification values of all categories are summed up; according to the cable fault category, the displacement influence is numerically quantified, and the displacement quantification values of all categories are summed up. By comparing the sizes of the summed values, the weight of the node influencing factor can be determined, and then a reasonable optimal weight value can be obtained.
[0067] like Figure 5 As shown, in one embodiment, the step S4 specifically includes:
[0068] Step S41, performing neighborhood difference analysis on the fused data set through a cable convolutional neural network to facilitate determining suspicious monitoring nodes;
[0069] Among them, neighborhood difference analysis means to perform domain analysis on the changes of each data in the fused data set. The initial analysis window of the domain analysis can be selected at will.
[0070] In addition, the initial analysis window can be limited according to the weight of the monitoring node (such as strain weight); suspicious monitoring nodes can be obtained after domain analysis.
[0071] Step S42, performing deep feature extraction on the current harmonic data of the cable affected by the suspicious monitoring node;
[0072] Wherein, the step S42 specifically includes:
[0073] Step S421, obtaining the historical current harmonic data of the cable in multiple fault states, and constructing a fusion feature space;
[0074] Specifically, for the acquired historical data, a variety of multi-wavelet perception kernels of different scales and types are constructed to perform full-band multi-scale and multi-level decomposition on the current harmonic data to obtain multiple frequency band components of the current harmonic data from low frequency to high frequency;
[0075] Each frequency band component is segmented using the smooth moving sampling method, and the information entropy of the lifted multi-wavelet is calculated as the original feature space;
[0076] Integrate the corresponding frequency band characteristic indicators of the original feature space of all frequency band components to obtain the low-frequency band cluster space and the high-frequency band cluster space;
[0077] The feature dimension reduction method is used to fuse the low-frequency band cluster space and the high-frequency band cluster space of the corresponding frequency band components to obtain a fused feature space from low frequency to high frequency.
[0078] Step S422, dividing the fused feature space into a training feature space and a test feature space;
[0079] Step S423: performing deep feature mining on the training feature space through a deep learning model;
[0080] In an example of this step, the implementation of deep feature mining can be to set a deep feature mining module in the deep learning model to mine deep features;
[0081] In this example, the deep feature mining module may be composed of two convolutional layers, two pooling layers, and one fully connected layer; the specific scales of the convolutional layers and the pooling layers may be set as needed.
[0082] Step S424: performing type recognition on the mined deep features and outputting recognition results;
[0083] Step S425: input the test feature space to judge the recognition result and determine the optimal deep learning model;
[0084] Step S426: Perform deep feature extraction on the current harmonic data using the optimal deep learning model. Figure 6 shown.
[0085] Step S43, classifying the cable degradation according to the extracted depth features;
[0086] Step S44: output the cable monitoring results according to the classification.
[0087] In one embodiment, Figure 7 As shown, the method also includes: step S5, training the cable convolutional neural network.
[0088] In an example of this embodiment, Figure 8 As shown, the step S5 specifically includes:
[0089] Step S51: construct a sample set and divide it into a training set and a validation set according to a certain ratio;
[0090] The certain ratio is initially set to 7:3, but of course, it can also be other ratios.
[0091] Step S52: Use the training set and the validation set to train the convolutional neural network model, optimize the model parameters and structure, and obtain a cable convolutional neural network that is adapted to the characteristics of the fusion data set.
[0092] In an example of this embodiment, Fig. 9 As shown, the method further includes: step S6, periodically adjusting the preset position of the monitoring node;
[0093] Among them, when the cable is monitored to be in normal operation, the preset position of the monitoring node can be adjusted, which helps to find hidden dangers that may be missed outside the monitoring node;
[0094] The adjustment cycle can be one quarter, one month, half a year, etc.
[0095] The adjusted preset positions may be equidistant, or gradually increase from the initial positions of the monitoring nodes, but are not limited thereto.
[0096] like Fig.10 As shown, in one embodiment, a cable monitoring system based on data fusion is used for the cable monitoring method based on data fusion, and the system includes:
[0097] The first database module 100 is used to obtain the current harmonic data of the cable through a preset monitoring node for monitoring the cable, and establish a current harmonic database;
[0098] The second database module 200 is used to obtain the strain data of the cable at the monitoring node and the displacement change data of the monitoring node, and establish a node influence factor library;
[0099] The data fusion module 300 is used to align the current harmonic database and the node influence factor database according to the acquisition time and the monitoring node position, and pre-process the aligned fusion data set;
[0100] The cable classification module 400 is used to perform degradation classification on the fused data set through the trained cable convolutional neural network, and output the cable monitoring results according to the classification results.
[0101] The system of this embodiment, wherein the cable convolutional neural network is capable of effectively identifying and classifying thermal aging, outer sheath damage, shielding layer damage, and excessive bending, can provide users with appropriate decision-making basis for taking preventive measures in making timely repairs and replacements of cables, thereby avoiding adverse consequences caused by failures.
[0102] like Fig.11 As shown, in one embodiment, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, steps S1 to S4 of the cable monitoring method based on data fusion are implemented.
[0103] The present embodiment provides a cable monitoring method based on data fusion, and based on the cable monitoring method based on data fusion, provides a cable monitoring system based on data fusion. The cable monitoring method based on data fusion is non-invasive monitoring, which can realize end-to-end prediction without prior knowledge, and can overcome the defects of not being able to conduct on-site live detection, or the monitoring parameters obtained are relatively single, and the damage and deterioration of the cable cannot be accurately assessed; select appropriate monitoring nodes according to the actual laying of the cable, for example: select the location or vicinity of the cable connection joint or connector; the current harmonic data of the cable can be monitored by instruments such as a harmonic analyzer, the strain data of the cable can be obtained by a tension sensor, and the displacement change data of the monitoring node can be obtained by sensors such as a displacement sensor or a gyroscope; after obtaining relevant data through each sensor, the data can be normalized first to facilitate data fusion; thereafter, feature extraction is performed on the fused data set, and the cables are classified according to the extracted features, and the classification results are saved, so as to quantitatively display the cable monitoring and guide the operation and maintenance of the cable, and can also solve the problems of single, insufficient and poor universality of parameters in the prior art when detecting cables.
[0104] Fig.11The internal structure diagram of a computer device in one embodiment is shown. The computer device includes a processor, a memory, a network interface, an input device and a display screen connected through a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement a cable monitoring method based on data fusion. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can execute a cable monitoring method based on data fusion. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.
[0105] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.
[0106] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A cable monitoring method based on data fusion, characterized in that: The method comprises: Step S1, obtaining the current harmonic data of the cable through a preset monitoring node for monitoring the cable, and establishing a current harmonic database; Step S2, obtaining the strain data of the cable at the monitoring node and the displacement change data of the monitoring node, and establishing a node influence factor library; Step S3, aligning the current harmonic database and the node influence factor database according to the acquisition time and the monitoring node position, and preprocessing the aligned fusion data set; Step S4: classify the degradation of the fused data set through the trained cable convolutional neural network, and output the cable monitoring results according to the classification.
2. The cable monitoring method based on data fusion according to claim 1 is characterized in that: The step S1 comprises: Step S11, after obtaining the current harmonic data of the cable, determine the fundamental wave of the current harmonic of the cable; Step S12, after shifting the basis function of the fundamental wave according to the wavelet transform formula, perform inner product with the signal to be analyzed at different scales to obtain the distribution of higher harmonics; Step S13: Establish a current harmonic database according to the distribution of each high-order harmonic.
3. The cable monitoring method based on data fusion according to claim 1 is characterized in that: In the step S2, based on the distributed communication network, the strain data of the cable at the monitoring node and the displacement change data of the monitoring node are obtained.
4. The cable monitoring method based on data fusion according to claim 3 is characterized in that: The method further includes: step S210, generating a maintenance request, specifically including: Step S211, gridding the distributed communication network and monitoring nodes; Step S212: judging whether the monitoring node is available according to the communication status of each communication module in the distributed communication network; Step S213: When the monitoring node is unavailable, a maintenance request is fed back; the maintenance request includes the device ID and setting address of the communication module corresponding to the monitoring node.
5. The cable monitoring method based on data fusion according to claim 4 is characterized in that: In step S211, after the distributed communication network and monitoring nodes are gridded, According to the gridded monitoring nodes, a weight setting directory is established and associated with the cable fault database; Based on the weight setting directory, the cable fault classification data in the cable fault database is called to determine the strain weight of the cable and the displacement weight of the monitoring node: After assigning the strain weight and displacement weight, the monitoring nodes are iterated to form a node change field; The optimal values of strain weight and displacement weight are selected according to the node change field, and the node influence factor library is updated.
6. The cable monitoring method based on data fusion according to claim 1, characterized in that: The step S4 specifically includes: Step S41, performing neighborhood difference analysis on the fused data set through a cable convolutional neural network to facilitate determining suspicious monitoring nodes; Step S42, performing deep feature extraction on the current harmonic data of the cable affected by the suspicious monitoring node; Step S43, classifying the cable degradation according to the extracted depth features; Step S44: output the cable monitoring results according to the classification.
7. The cable monitoring method based on data fusion according to claim 6 is characterized in that: The step S42 specifically includes: Step S421, obtaining the historical current harmonic data of the cable in multiple fault states, and constructing a fusion feature space; Step S422, dividing the fused feature space into a training feature space and a test feature space; Step S423: performing deep feature mining on the training feature space through a deep learning model; Step S424: performing type recognition on the mined deep features and outputting recognition results; Step S425: input the test feature space to judge the recognition result and determine the optimal deep learning model; Step S426: Perform deep feature extraction on the current harmonic data through the optimal deep learning model.
8. The cable monitoring method based on data fusion according to claim 1, characterized in that: The method also includes: step S5, training the cable convolutional neural network.
9. The cable monitoring method based on data fusion according to claim 8, characterized in that: The step S5 specifically includes: Step S51: construct a sample set and divide it into a training set and a validation set according to a certain ratio; Step S52: Use the training set and the validation set to train the convolutional neural network model, optimize the model parameters and structure, and obtain a cable convolutional neural network that is adapted to the characteristics of the fusion data set.
10. The cable monitoring method based on data fusion according to claim 1, characterized in that: The method further includes: step S6, periodically adjusting the preset position of the monitoring node.
