Cable system fault analysis method based on big data

Abstract

The invention discloses a cable system fault analysis method based on big data, and belongs to the technical field of cable fault analysis, and the method specifically comprises the steps: implanting flexible deformation sensing units at a plurality of preset position points of a cable, and enabling each sensing unit to synchronously collect the multi-dimensional deformation data of the position point; mapping the real-time deformation data acquired by each sensing unit to the virtual cable three-dimensional model to generate a sensing node; performing spatial-temporal clustering analysis on the deformation data of all the sensing nodes, calculating deformation feature similarity measurement values of different sensing node data, and marking the sensing nodes with abnormal data modes as an abnormal node set; according to spatial distribution characteristics of the abnormal node set, a cable fault probability interval section is positioned by analyzing a diffusion direction of an abnormal data cluster and deformation data correlation of adjacent nodes; a visual warning mark is marked; therefore, the defect detection of the early failure of non-electrical factors is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of cable fault analysis, and in particular to a cable system fault analysis method based on big data. Background Art

[0002] As modern society becomes increasingly dependent on electricity, the stability and reliability of cable operation, as a key carrier of power transmission, are of vital importance. In the construction of large-scale power infrastructure such as smart grids and urban integrated pipeline corridors, long-distance, large-capacity cable systems have been widely used. However, the cable laying environment is complex and is affected by geological changes, external force damage, temperature changes, material aging and other factors. It is easy to fail, which in turn causes large-scale power outages, bringing serious impacts on social economy and people's lives. For a long time, cable fault detection and analysis methods have been centered around electrical parameters, with monitoring of parameters such as current, voltage, and resistance becoming the core of traditional methods. Although this monitoring mode can detect obvious electrical faults to a certain extent, it is powerless against early faults caused by non-electrical factors such as cable stretching and extrusion caused by mechanical stress that causes local cable structure deformation, or geological settlement. For example, when the cable begins to show slight structural deformation due to mechanical stress generated by surrounding construction excavation, its electrical parameters may remain within the normal fluctuation range for quite some time, making it difficult to accurately reflect this potential fault in a timely manner. By the time the electrical parameters change significantly, the cable fault has often developed to a more serious stage.

[0003] Therefore, the existing method of monitoring cable faults that only relies on electrical parameters has serious monitoring blind spots when dealing with early faults caused by non-electrical factors in complex operating environments, making it difficult to achieve early and accurate warning and effective diagnosis of cable faults. Summary of the invention

[0004] The purpose of the present invention is to provide a method for analyzing a fault in a large data cable system to solve the following technical problems: Only collecting parameters in the electrical dimension makes it impossible to analyze early failures caused by non-electrical factors such as mechanical stress and structural deformation.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A cable system fault analysis method based on big data, comprising the following steps: Flexible deformation sensing units are implanted at several preset positions of the cable, and each sensing unit synchronously collects multi-dimensional deformation data of the position point, including surface curvature change, axial tensile strain rate and radial compression fluctuation frequency; Map the real-time deformation data collected by each sensing unit to a virtual 3D cable model to generate sensing nodes, and dynamically adjust the model weights based on the deformation coupling relationship between adjacent sensing nodes; Perform spatio-temporal clustering analysis on the deformation data of all sensing nodes, calculate the similarity metric values of the deformation characteristics of the data of different sensing nodes, and mark the sensing nodes with abnormal data patterns as an abnormal node set; According to the spatial distribution characteristics of the abnormal node set, locate the cable fault probability interval section by analyzing the diffusion direction of the abnormal data cluster and the correlation of the deformation data of adjacent nodes; Mark a visual warning sign on the corresponding physical actual interval section of the cable in the cable fault probability interval section.

[0006] As a further solution of the present invention: the flexible deformation sensing unit is composed of a highly elastic base material and a distributed optical fiber sensing array. The surface of the highly elastic base material is provided with a wavy microstructure, and a flexibly wound flexible circuit is embedded inside. The optical fiber sensing array covers the circumferential surface of the cable in a staggered arrangement manner, and the communication link between sensing units adopts an adaptive frequency hopping mechanism.

[0007] As a further solution of the present invention: the construction process of the virtual 3D cable model is as follows: Discretize the physical structure of the cable into a weighted node network. The node weight reflects the historical fault probability of the corresponding position point. Input the real-time deformation data into the node network, dynamically update the connection edge weights according to the deformation data difference degree of adjacent nodes, introduce a time decay factor to adjust the data weights of deformation data in different time sequences, and automatically trigger the topology reconstruction mechanism when it is detected that the weights of three consecutive nodes increase synchronously and abnormally.

[0008] As a further solution of the present invention: the specific process of the spatio-temporal clustering analysis is as follows: Convert the deformation data collected by each sensing node into a data unit containing spatial position coding, time series identification, and multi-dimensional feature vectors, identify abnormal data patterns through a preset spatio-temporal density clustering algorithm. In the spatio-temporal density clustering algorithm, cable physical structure constraint conditions are introduced, set spatial correlation weights are assigned to nodes in the same material characteristic section, and the conduction attenuation characteristics of the deformation characteristics along the cable axis are excluded synchronously during the clustering process. When abnormal data of an isolated node is detected, distinguish the real fault signal from local interference noise by comparing the deformation fluctuation correlation of the upstream and downstream nodes of the isolated node within a preset time window, and finally generate a set of abnormal data clusters with spatio-temporal continuity. The spatial distribution of the data cluster set represents the interval range of potential cable faults.

[0009] As a further solution of the present invention: the process of the spatio-temporal clustering analysis further includes: The generated abnormal data cluster set is verified for physical rationality, and the cable material fatigue characteristics database is called to screen out data clusters that conform to the material damage evolution law. A dynamic weight adjustment mechanism is established to automatically adjust the clustering density threshold according to the time continuity and spatial extensibility of the data cluster. An expert experience rule base is introduced to perform confidence weighting on data clusters located at cable branch nodes or historical fault-prone areas. Finally, the verified optimized clustering results are output, and the fault type labels are generated by comparing the clustering feature vectors with the feature templates in the typical fault mode database.

[0010] As a further solution of the present invention: the process of locating the cable fault probability interval segment includes: Taking the core node of the abnormal data cluster as the base point, a dynamic detection interval is established by extending to both sides. The interval boundary is determined by analyzing the gradient change rate and statistical distribution consistency of the deformation data in the extension direction. For the interval segment crossing the cable branch node, the branch line historical fault database is automatically called to adjust the boundary threshold, generate a preliminary fault probability interval segment and attach a confidence rating.

[0011] As a further solution of the present invention: the calculation process of the confidence rating is: The deformation feature vector within the preliminary fault probability interval is extracted and matched with the feature template in the typical fault mode database for similarity. The probability value of the fault occurrence in the interval is calculated based on the matching result. At the same time, the time dimension analysis is introduced to assign a higher confidence weight to the interval where abnormal data continues to appear. Finally, the optimized fault probability interval with a confidence rating is output. The confidence rating is used to mark the priority sorting of operation and maintenance.

[0012] As a further solution of the present invention: the visual warning sign marks the fault probability interval segment in the form of a gradient color band, and the color depth of the color band is directly proportional to the density of abnormal data aggregation in the interval.

[0013] Beneficial effects of the present invention: By implanting a flexible deformation sensing unit at the preset position of the cable that can collect multi-dimensional deformation data such as surface curvature change, axial tensile strain rate, and radial compression fluctuation frequency, the defects of electrical parameter monitoring alone that cannot reflect the physical deformation state of the cable and are difficult to detect early failures caused by non-electrical factors are compensated. By mapping real-time deformation data to a virtual cable three-dimensional model and dynamically adjusting the model weights, combined with spatiotemporal clustering analysis, abnormal nodes and fault probability intervals can be accurately located, overcoming the problems of incomplete monitoring and inaccurate fault diagnosis of traditional methods. The visual warning sign intuitively presents the fault probability interval, and the color depth of the color band is proportional to the density of abnormal data aggregation, achieving timely warning and ensuring the safe and stable operation of the cable system. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below in conjunction with the accompanying drawings.

[0015] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] See also Figure 1 As shown, the present invention is a method for analyzing a fault in a cable system based on big data, comprising the following steps: First, flexible deformation sensing units are cleverly implanted at several precisely preset locations of the cable. These sensing units can be called "pioneers of intelligent perception". They are composed of a composite of a highly elastic base material and a distributed optical fiber sensor array. The surface of the highly elastic base material is provided with a unique wavy microstructure, and a spirally wound flexible circuit is embedded inside. The optical fiber sensor array tightly covers the circumferential surface of the cable in an interlaced arrangement. The communication link between the sensing units adopts an adaptive frequency hopping mechanism to ensure stable data transmission. Each sensing unit can synchronously collect rich multi-dimensional deformation data at its location, including the surface curvature change that accurately reflects the change in the degree of bending of the cable surface, the axial tensile strain rate that clearly reflects the degree of stretching of the cable in the axial direction, and the radial compression fluctuation frequency that accurately represents the change in the frequency of compression of the cable in the radial direction.

[0018] Subsequently, the real-time deformation data collected by each sensor unit is mapped to a carefully constructed virtual cable three-dimensional model using a sophisticated algorithm to generate sensor nodes. In this virtual model, the model weights are dynamically adjusted based on the deformation coupling relationship between adjacent sensor nodes, that is, when a node is deformed, it will affect the adjacent nodes. For example, when a node undergoes axial tensile strain due to external factors, the strain of the adjacent nodes will be fed back to the model, and the model can be more closely aligned with the actual operating status of the cable by continuously adjusting the weights.

[0019] Next, an in-depth spatiotemporal clustering analysis is conducted on the deformation data of all sensor nodes. The deformation data collected by each sensor node is first converted into a data unit containing spatial position coding, time series identification and multi-dimensional feature vectors, and then the abnormal data pattern is identified with the help of a preset spatiotemporal density clustering algorithm. In this process, the physical structure constraints of the cable are introduced, and the nodes in the same material property segment are given a set spatial correlation weight, while the conductive attenuation characteristics of the deformation characteristics along the cable axis are cleverly excluded. When the abnormal data of an isolated node is detected, by comparing the deformation fluctuation correlation of the upstream and downstream nodes of the isolated node within the preset time window, the real fault signal and the local interference noise are accurately distinguished, and finally a set of abnormal data clusters with spatiotemporal continuity is generated. The spatial distribution of these data clusters intuitively represents the range of potential cable faults.

[0020] Then, according to the spatial distribution characteristics of the abnormal node set, the core node of the abnormal data cluster is used as the base point to extend to both sides to establish a dynamic detection interval. The interval boundary is determined by carefully analyzing the gradient change rate and statistical distribution consistency of the deformation data in the extension direction. For the interval segment crossing the cable branch node, the branch line historical fault database is automatically called to adjust the boundary threshold, so as to accurately locate the cable fault probability interval segment and add a confidence rating.

[0021] Finally, in the actual physical interval of the cable corresponding to the cable fault probability interval, a visual warning sign is marked in an intuitive and eye-catching way. The sign is presented in the form of a gradient color band, and the color depth of the color band is proportional to the density of abnormal data in the interval. The staff can quickly judge the probability of fault by the color depth, so as to take corresponding measures in time, which greatly ensures the safe and stable operation of the cable system.

[0022] In a preferred embodiment of the present invention, the flexible deformation sensing unit used has a unique and exquisite structure. It is composed of a high elastic base material and a distributed optical fiber sensor array, and this combination is a perfect match. The high elastic base material is like a solid and resilient "protective armor", providing basic support for the entire sensing unit. The wavy microstructure carefully arranged on its surface greatly increases the contact friction with the cable surface, ensuring that the sensing unit is firmly attached to the cable surface and is not easily displaced due to external impact or daily vibration of the cable. For example, in some urban underground pipe corridors, the cable will produce a small displacement due to the vibration of passing vehicles, and the wavy microstructure can effectively deal with such situations. The spirally wound flexible circuit embedded inside is like the human body's neural network, responsible for transmitting various electrical signals to ensure the stability of data collection and transmission. The optical fiber sensor array tightly covers the circumferential surface of the cable in an interlaced arrangement, and can sense the subtle deformation of the cable in all directions in an all-round and dead-angle manner. Optical fiber sensing units at different positions can simultaneously monitor the deformation of different parts of the cable, just like multiple "scouts" working together to ensure comprehensive information acquisition. The communication link between the sensor units uses an adaptive frequency hopping mechanism, which can automatically adjust the communication frequency according to changes in the surrounding electromagnetic environment, effectively avoiding signal interference. For example, near a substation with a complex electromagnetic environment, the adaptive frequency hopping mechanism allows the sensor units to transmit data stably and ensure the accuracy of monitoring.

[0023] In another preferred embodiment of the present invention, the construction process of the virtual cable three-dimensional model contains advanced technical logic. First, the physical structure of the cable is discretized into a node network with weights, which is like dividing a continuous cable into multiple key "monitoring points". Each node weight reflects the historical failure probability of the corresponding position point. For example, in the past operation, a certain section of cable was frequently affected by the vibration of construction machinery due to its proximity to the construction site, resulting in frequent failures. Then, the node weight corresponding to this position will be relatively high. Then, the real-time deformation data is input into this node network, and the connection edge weight is dynamically updated according to the difference of the deformation data of adjacent nodes. Assuming that the adjacent nodes A and B, the A node detects a sudden increase in the axial tensile strain rate, while the strain rate change of the B node is relatively small. At this time, the edge weight connecting the A and B nodes will be adjusted according to this difference to more accurately reflect the actual mechanical conduction relationship of the cable. At the same time, the time decay factor is introduced to adjust the data weight of the deformation data of different time sequences, and a higher weight is given to the deformation data that occurred recently, because it can better reflect the current operation status of the cable. For example, the weight of the deformation data a week ago will be lower than the newly collected data yesterday. When the weights of three consecutive nodes are detected to have grown abnormally simultaneously, the topology reconstruction mechanism is automatically triggered. For example, if the weights of three adjacent nodes in a certain area have grown significantly in a short period of time due to geological settlement, the system will immediately start topology reconstruction and re-optimize the node network to more accurately monitor the fault hazards of the cables in the area and provide a more reliable model basis for subsequent fault analysis.

[0024] In another preferred embodiment of the present invention, the specific process of spatiotemporal clustering analysis is: First, the deformation data collected by each sensor node is processed and converted into a data unit containing spatial position coding, time series identification and multi-dimensional feature vector. Spatial position coding is like giving each sensor node a unique "geographic coordinate" to clearly identify its specific location on the cable. For example, if the cable is divided into multiple sections, and each section has multiple monitoring points, the spatial position coding can be accurate to the specific section and monitoring point number. The time series identification records the time information of data collection, which can reflect the change of deformation data over time, such as marking the deformation data of a monitoring point at different times of the day. The multi-dimensional feature vector contains multi-dimensional information such as surface curvature change, axial tensile strain rate and radial compression fluctuation frequency, which comprehensively describes the deformation state of the cable at that position.

[0025] Next, a preset spatiotemporal density clustering algorithm is used to identify abnormal data patterns. The algorithm introduces cable physical structure constraints because the physical structure and material properties of different parts of the cable are different, which will affect the propagation and performance of deformation. For example, when the cable passes through different geological areas, the stress on the cable will be different due to the different hardness and stability of the soil. Therefore, the set spatial association weight will be given to the nodes in the same material property segment. Assuming that a section of cable is laid in a uniform rock layer, the spatial association weight between the nodes in this section is high because their deformations may affect each other. At the same time, the conductive attenuation characteristics of the deformation features along the axial direction of the cable should be excluded in the clustering process. When the cable transmits deformation in the axial direction, it will attenuate with the increase of distance, just like the sound will gradually weaken during the propagation process. Excluding this feature can avoid misjudgment caused by attenuation and make the clustering results more accurate.

[0026] When abnormal data of an isolated node is detected, it is necessary to carefully distinguish whether it is a real fault signal or local interference noise. The judgment is made by comparing the deformation fluctuation correlation of the upstream and downstream nodes of the isolated node within the preset time window. For example, within a one-hour time window, if an isolated node has an abnormal axial tensile strain rate, but its upstream and downstream nodes have normal deformation fluctuations at the same time and no obvious correlation, then this abnormal data is likely to be local interference noise; conversely, if the upstream and downstream nodes also show similar deformation fluctuation trends, it is more likely to be a real fault signal. After such a series of processing, a set of abnormal data clusters with spatiotemporal continuity is finally generated. The spatial distribution of these data clusters clearly characterizes the range of potential cable faults.

[0027] In a preferred case of this embodiment, in order to further improve the accuracy and reliability of the spatiotemporal clustering analysis, the generated abnormal data cluster set is optimized in many aspects.

[0028] The first step is to verify the physical rationality. By calling the cable material fatigue characteristics database, we can screen out data clusters that conform to the damage evolution law of the material. Different cable materials will have specific damage evolution laws during long-term use. For example, a certain cable material will have a specific deformation pattern after a certain number of stretching and compression. If an abnormal data cluster conforms to the damage evolution law of this material, then it is more likely to represent a real fault.

[0029] The second step is to establish a dynamic weight adjustment mechanism. The cluster density threshold is automatically adjusted according to the time continuity and spatial extension of the data cluster. If a data cluster continues to have abnormal data, it means that the fault may be developing; if it has a large extension range in space, it also means that the fault may be serious. In this case, appropriately lowering the cluster density threshold can capture fault information more sensitively. For example, when a data cluster has abnormalities for three consecutive days and covers a long distance of the cable, the cluster density threshold can be lowered so that more relevant abnormal data can be included in the data cluster.

[0030] The third step is to introduce the expert experience rule base. Confidence weighting is performed on data clusters located at cable branch nodes or historical high-prone areas. The stress distribution at cable branch nodes is relatively complex and prone to failure; historical high-prone areas have a higher probability of failure occurring again. Experts have summarized some rules based on past experience. For example, at a certain branch node, if a certain type of deformation anomaly occurs, the probability of failure will be higher. By introducing the expert experience rule base, data clusters in these areas are given higher confidence weights, which can more accurately assess the risk of failure.

[0031] Finally, the verified optimized clustering results are output, and the fault type label is generated by comparing the cluster feature vector with the feature template in the typical fault mode database. The typical fault mode database stores feature templates of various common faults, such as partial damage of the cable, excessive stretching and other faults, which have their own unique characteristics. By comparing the cluster feature vector with these templates, the type of fault can be quickly and accurately determined, providing a strong basis for subsequent maintenance and processing.

[0032] In another preferred embodiment of the present invention, the process of locating the cable fault probability interval segment includes: First, the core node of the abnormal data cluster is used as the base point. The core node of the abnormal data cluster is determined in the previous spatiotemporal cluster analysis. It represents the location where the abnormal data is most concentrated and the characteristics are most obvious, just like the center of a storm. Starting from this core node, a dynamic detection interval is established on both sides. This dynamic detection interval is not fixed, it will be adjusted according to the actual situation of the cable and the changes in data.

[0033] When determining the interval boundary, it is necessary to analyze the gradient change rate and statistical distribution consistency of the deformation data in the extension direction. The gradient change rate of the deformation data reflects the speed of the deformation change in space. For example, if the gradient change rate of the deformation data suddenly increases in a certain direction, it means that the deformation of the cable in this direction has changed drastically, which is likely to be the boundary affected by the fault. Statistical distribution consistency refers to analyzing whether the data in the extension direction conforms to a certain specific statistical law. If the statistical distribution of the data shows obvious differences in a certain position, it can also be used as a basis for determining the interval boundary.

[0034] The situation becomes more complicated when encountering a section that crosses a cable branch node. A cable branch node is a critical location in a cable line. Its structure and stress conditions are different from those of an ordinary line, and it has historically been more prone to failures. At this point, the system automatically calls the branch line historical failure database. This database records various information about past failures of the branch line, including the location, type, and frequency of the failures. By analyzing this historical data, the boundary threshold can be adjusted. For example, if a branch node has often experienced failures due to external force squeezing in the past, when determining the boundary of the section that crosses the node, the boundary range will be appropriately expanded to more comprehensively cover the area where the failure may occur. After this processing, a preliminary failure probability section is generated, and a confidence rating is attached to it.

[0035] The confidence rating calculation process is a further evaluation of the initial failure probability interval. It can help operation and maintenance personnel more accurately determine the possibility and severity of the failure. The process is as follows: The first step is to extract the deformation feature vector within the initial fault probability interval. The deformation feature vector contains various deformation information of the cable within the interval, such as the change in surface curvature, axial tensile strain rate, and radial compression fluctuation frequency. These feature vectors are matched with the feature templates in the typical fault mode database for similarity. The typical fault mode database stores feature templates of various known faults, which is like a "dictionary" of faults. For example, if the deformation feature vector within a certain interval is very similar to the feature template of "partial cable damage" in the database, it means that the probability of partial cable damage in this interval is relatively high. Based on the matching results, the probability value of the fault in the interval is calculated.

[0036] At the same time, the time dimension analysis is introduced. If an interval segment continues to have abnormal data, it means that the fault may be developing or has existed for a period of time, and its severity may be higher. Therefore, a higher confidence weight will be given to the interval segment that continues to have abnormal data. For example, if an interval segment detects abnormal axial tensile strain rate for three consecutive days, then its confidence weight will be higher than that of an interval segment that only has abnormal data in one day.

[0037] Finally, the optimized fault probability interval with confidence rating is output. This confidence rating is like a "priority label" that operation and maintenance personnel can use to determine the order of operation and maintenance. Intervals with high confidence ratings mean that the possibility of failure is greater and the impact is more serious, and they need to be inspected and maintained first; while intervals with low confidence ratings can be processed later. This can improve the efficiency of operation and maintenance work and ensure that limited resources can be used reasonably.

[0038] In another preferred embodiment of the present invention, the visual warning sign uses a gradient color band to intuitively mark the fault probability interval of the cable. The higher the density of abnormal data aggregation in the interval, the deeper the color depth of the color band. For example, in a certain section of cable, if the abnormal data is densely distributed due to long-term external force compression, the color of the color band corresponding to this interval will be dark and rich; while in areas where the abnormal data is relatively sparse, the color of the color band will be light. Operation and maintenance personnel can quickly lock in high-risk fault areas by naked eye observation alone, greatly improving the efficiency of troubleshooting and ensuring the stable operation of the cable system.

[0039] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A cable system fault analysis method based on big data, characterized in that: The following steps are involved: Flexible deformation sensing units are implanted at several preset positions of the cable, and each sensing unit synchronously collects multi-dimensional deformation data of the position point, including surface curvature change, axial tensile strain rate and radial compression fluctuation frequency; The real-time deformation data collected by each sensor unit is mapped to the virtual cable 3D model to generate sensor nodes, and the model weight is dynamically adjusted based on the deformation coupling relationship between adjacent sensor nodes; Perform spatiotemporal clustering analysis on the deformation data of all sensor nodes, calculate the deformation feature similarity measurement values ​​of different sensor node data, and mark the sensor nodes with abnormal data patterns as abnormal node sets; According to the spatial distribution characteristics of the abnormal node set, the cable fault probability interval is located by analyzing the diffusion direction of the abnormal data cluster and the correlation of the deformation data of the adjacent nodes; A visual warning sign is marked on the actual physical interval of the cable corresponding to the cable fault probability interval.

2. A method for analyzing cable system failure based on big data according to claim 1, characterized in that: The flexible deformation sensing unit is composed of a high-elastic base material and a distributed optical fiber sensor array. A wavy microstructure is arranged on the surface of the high-elastic base material, and a spirally wound flexible circuit is embedded inside. The optical fiber sensor array covers the circumferential surface of the cable in an interlaced arrangement, and the communication link between the sensor units adopts an adaptive frequency hopping mechanism.

3. A method for analyzing cable system failure based on big data according to claim 1, characterized in that: The construction process of the virtual cable three-dimensional model is as follows: The physical structure of the cable is discretized into a weighted node network. The node weight reflects the historical failure probability of the corresponding location point. The real-time deformation data is input into the node network. The connection edge weight is dynamically updated according to the difference between the deformation data of adjacent nodes. The time attenuation factor is introduced to adjust the data weight of deformation data in different time sequences. When it is detected that the weights of three consecutive nodes grow abnormally synchronously, the topology reconstruction mechanism is automatically triggered.

4. A method for analyzing cable system failure based on big data according to claim 1, characterized in that: The specific process of the spatiotemporal clustering analysis is as follows: The deformation data collected by each sensor node is converted into a data unit containing spatial position coding, time series identification and multi-dimensional feature vectors. Abnormal data patterns are identified by a preset spatiotemporal density clustering algorithm. The physical structure constraints of the cable are introduced into the spatiotemporal density clustering algorithm. The nodes in the same material property segment are assigned set spatial correlation weights. In the clustering process, the conduction attenuation characteristics of the deformation characteristics along the cable axis are simultaneously excluded. When abnormal data of an isolated node is detected, the deformation fluctuation correlation of the upstream and downstream nodes of the isolated node within a preset time window is compared to distinguish between real fault signals and local interference noise. Finally, a set of abnormal data clusters with spatiotemporal continuity is generated. The spatial distribution of the data cluster set represents the interval range of potential cable faults.

5. A method for analyzing a cable system fault based on big data according to claim 4, characterized in that: The process of the spatiotemporal clustering analysis also includes: The generated abnormal data cluster set is verified for physical rationality, and the cable material fatigue characteristics database is called to screen out data clusters that conform to the material damage evolution law. A dynamic weight adjustment mechanism is established to automatically adjust the clustering density threshold according to the time continuity and spatial extensibility of the data cluster. An expert experience rule base is introduced to perform confidence weighting on data clusters located at cable branch nodes or historical fault-prone areas. Finally, the verified optimized clustering results are output, and the fault type labels are generated by comparing the clustering feature vectors with the feature templates in the typical fault mode database.

6. A method for analyzing cable system failure based on big data according to claim 1, characterized in that: The process of locating the cable fault probability interval segment includes: Taking the core node of the abnormal data cluster as the base point, a dynamic detection interval is established by extending to both sides. The interval boundary is determined by analyzing the gradient change rate and statistical distribution consistency of the deformation data in the extension direction. For the interval segment crossing the cable branch node, the branch line historical fault database is automatically called to adjust the boundary threshold, generate a preliminary fault probability interval segment and attach a confidence rating.

7. A method for analyzing cable system failure based on big data according to claim 6, characterized in that: The confidence rating is calculated as follows: The deformation feature vector within the preliminary fault probability interval is extracted and matched with the feature template in the typical fault mode database for similarity. The probability value of the fault occurrence in the interval is calculated based on the matching result. At the same time, the time dimension analysis is introduced to assign a higher confidence weight to the interval where abnormal data continues to appear. Finally, the optimized fault probability interval with a confidence rating is output. The confidence rating is used to mark the priority sorting of operation and maintenance.

8. The method for analyzing cable system failure based on big data according to claim 1, characterized in that: The visual warning sign marks the fault probability interval segment in the form of a gradient color band, and the color depth of the color band is directly proportional to the density of abnormal data aggregation in the interval.

Images