Cable safety monitoring system and method based on underground unmanned aerial vehicle autonomous inspection

Abstract

The invention discloses a cable safety monitoring system and method based on autonomous inspection of an underground unmanned aerial vehicle, and relates to the technical field of inspection control. A cable safety monitoring area is set, and a cable monitoring topological graph is constructed based on the position of each monitoring node in the set cable safety monitoring area; arranging an unmanned aerial vehicle to collect cable monitoring data in real time based on the constructed cable monitoring topological graph; and meanwhile, collecting cable historical monitoring data in real time and constructing a sample set, training the cable historical monitoring data in the sample set to obtain a trained safety monitoring model, finally summarizing the cable monitoring data collected by the unmanned aerial vehicle in real time, and performing analysis through the trained safety monitoring model. Whether the cable monitoring data collected in real time is abnormal or not is judged, after the abnormity is monitored, a worker is arranged for processing, and the accuracy of unmanned aerial vehicle inspection and monitoring is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of inspection control, and specifically relates to a cable safety monitoring system and method based on autonomous inspection of underground drones. Background Art

[0002] Currently, drones have been increasingly widely used in the fields of inspection and monitoring due to their small size, flexibility, and real-time detection capabilities. With the continuous expansion of the inspection scope, in order to more efficiently complete the inspection tasks and expand the monitoring scope, the inspection scheduling and inspection accuracy of drones have become crucial.

[0003] An existing technology, such as a drone inspection system disclosed in the invention patent application with the publication number CN113204247B, its method includes: a man-machine operation management platform, and at least one drone communicating with the drone operation management platform; in the present invention, when the drone receives an inspection instruction sent by the drone operation management platform, it reads the inspection target information included in the instruction; then, based on the current battery power and the inspection target information, it conducts an inspection path planning to obtain an initial inspection path; then, it adjusts the initial inspection path according to the current meteorological information to obtain an adjusted target inspection path; finally, it flies to the target inspection position according to the target inspection path, obtains the inspection scene model corresponding to the target inspection position, and conducts an inspection on the inspection object according to the inspection scene model.

[0004] For the above-mentioned solution, it can be seen that the current drone inspection mainly plans the flight path based on the starting position, ending position of the drone, as well as the position and height of the inspection object, ignoring the overall process of the inspection, and only inspecting a single route, which has certain limitations. Summary of the Invention

[0005] The purpose of the present invention is to provide a cable safety monitoring system and method based on autonomous inspection of underground drones, which solves the problem of efficiently completing inspection tasks and expanding the monitoring scope in the background art.

[0006] To solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a cable safety monitoring method based on autonomous inspection of underground drones, specifically including the following steps:

[0007] S1. Set the cable safety monitoring area, and based on the positions of each monitoring node within the set cable safety monitoring area, construct a cable monitoring topology map;

[0008] S2. Based on the constructed cable monitoring topology map, arrange for the drone to collect cable monitoring data in real time;

[0009] S21. Plan each cable line in the constructed cable monitoring topology map through a path planning algorithm to obtain the shortest path between each monitoring node;

[0010] S22. Collect the current UAV status data in real time and schedule the UAV based on the shortest path between each monitoring node obtained;

[0011] S23. After the scheduling is completed, arrange the corresponding UAV for inspection and collect the cable monitoring data in real time;

[0012] S3. Collect the cable historical monitoring data in real time and construct a sample set, and train the cable historical monitoring data in the sample set to obtain a trained safety monitoring model;

[0013] S4. Summarize the cable monitoring data collected by the UAV in real time and analyze it through the trained safety monitoring model to determine whether there is any abnormality in the cable monitoring data collected in real time;

[0014] S5. After detecting an abnormality, arrange for staff to handle it.

[0015] Preferably, the steps of setting the cable safety monitoring area and constructing a cable monitoring topology map based on the positions of each monitoring node in the set cable safety monitoring area include:

[0016] Set the cable safety monitoring area and collect the path data in the cable safety monitoring area in real time;

[0017] Set the intersection points of the paths in the cable safety monitoring area as monitoring nodes;

[0018] Set the set of path data in the cable safety monitoring area as: e = {e1, e2,..., e m};

[0019] Among them, e represents the set of path data in the cable safety monitoring area, and e m represents the mth path data in the cable safety monitoring area;

[0020] Set the path data e m to be composed of a triple {a m , z m , d m}, where a m represents the starting monitoring point of the path data, z m represents the ending monitoring point of the path data, and d m represents the distance of the path;

[0021] Construct a cable monitoring topology map based on the collected set of path data.

[0022] Preferably, the step of obtaining the shortest path between each monitoring node by planning each cable line in the constructed cable monitoring topology map through a path planning algorithm includes the following steps:

[0023] S211. Initialize a path matrix based on the constructed cable monitoring topology map, and set the data in the matrix as the distance of this path;

[0024] S212. Randomly select a monitoring point, and based on the selected monitoring point as the starting point, update the path matrix again. Set that if there is a direct path between the monitoring point and the starting point, then set the distance of this direct path as the corresponding updated matrix data;

[0025] S213. Select the path with the smallest matrix data, take the monitoring points on this path as a whole, and update the path matrix again until all monitoring points are traversed, and output the shortest path between each monitoring node;

[0026] S214. Save the shortest path between each output monitoring node, and remove the corresponding path in the constructed cable monitoring topology map. If there is only one path between a monitoring point and other monitoring points, then the corresponding monitoring point in the cable monitoring topology map is also removed. After removing the corresponding path, iteratively execute steps S212 - S213 until there is no path between each monitoring node, and output multiple groups of the shortest paths between each monitoring node saved.

[0027] Preferably, the step of collecting the current UAV state data in real time and scheduling the UAV based on the obtained shortest path between each monitoring node includes the following steps:

[0028] S221. Collect UAV state data in real time and calculate the inspection cost of the UAV;

[0029] S222. Schedule the UAV based on the calculated inspection cost of the UAV and multiple groups of the shortest paths between each monitoring node.

[0030] Preferably, the step of collecting the UAV state data in real time and calculating the inspection cost of the UAV includes the following steps:

[0031] Collecting UAV state data in real time includes: the battery power of the UAV and the working state of the UAV;

[0032] Setting the inspection cost of the UAV includes: the battery power of the UAV and the inspection time of the UAV;

[0033] Among them, the UAV inspection time = data feedback time + data processing time + network delay time;

[0034] The battery power of the UAV = flight power consumption + data processing power consumption + data feedback power consumption. The inspection cost formula of the UAV is as follows:

[0035]

[0036] Among them, Z j represents the inspection cost of the j-th UAV, and E j represents the power required for the j-th UAV to perform an inspection, and T j represents the inspection time using the j-th UAV, represents the energy consumption weight of the j-th UAV, represents the time weight of the j-th UAV.

[0037] Preferably, the scheduling of the UAV based on the calculated UAV inspection cost and the shortest paths between multiple groups of monitoring nodes includes the following steps:

[0038] S222. Schedule the UAV based on the calculated UAV inspection cost and the shortest paths between multiple groups of monitoring nodes;

[0039] S2221. Initialize the algorithm parameters and perform chromosome encoding on the shortest paths between multiple groups of monitoring nodes;

[0040] S2222. Construct an initial population by enabling the saturated transportation mode;

[0041] The saturated transportation mode means that a random UAV is arranged for inspection on the shortest path between each group of monitoring nodes;

[0042] S2223. Calculate the population fitness through the fitness function;

[0043] Set each individual to represent the shortest path between a group of monitoring nodes, and the shortest path between each group of monitoring nodes represents the path of the UAV after inspecting all monitoring points in the current inspection path;

[0044] Set the fitness function:

[0045] Among them, Fit(x) represents the fitness of the individual x, and Z represents the inspection cost of the UAV;

[0046] S2224. Perform the genetic operator selection operation;

[0047] Set the probability of an individual being selected as p(x i );

[0048]

[0049] Among them, I is the population size, and Fit(x i ) is the fitness of the i-th individual;

[0050] S2225. Output the optimal solution to the problem;

[0051] Set the maximum number of iterations Y. Set that when the chromosome coding output after Y iterations converges, terminate the hybrid genetic algorithm, and use the chromosome coding output in the last iteration as the inspection path for the corresponding scheduled unmanned aerial vehicle.

[0052] Preferably, the steps of collecting the cable historical monitoring data in real time and constructing a sample set, and training the cable historical monitoring data in the sample set to obtain a trained safety monitoring model include the following:

[0053] S31. Collect the cable historical monitoring data, summarize the collected cable historical monitoring data and evenly divide it into two groups, one group as the training group and one group as the verification group;

[0054] S32. Divide each cable historical monitoring data in the training group into n cable historical monitoring data blocks by an equal-proportion division method;

[0055] S33. Input the divided cable historical monitoring data blocks into the convolutional neural network and train them through the convolutional neural network;

[0056] S35. Input the cable historical monitoring data in the verification group into the convolutional neural network, and match the output result of the verification group with the output result of the training group;

[0057] Set a matching success threshold. When the matching result is greater than or equal to the set threshold, output the current convolutional neural network as the recognition model. When the matching result is less than the set threshold, adjust the weights of the convolutional neural network until the matching result is greater than the set threshold;

[0058] Set the adjusted convolutional neural network as the trained safety monitoring model.

[0059] Preferably, the steps of inputting the divided cable historical monitoring data blocks into the convolutional neural network and training them through the convolutional neural network include the following:

[0060] After the convolutional neural network receives the cable historical monitoring data block, set the convolutional kernel size and initial weights of the convolutional layer in the convolutional neural network, and calculate the data features of the corresponding cable historical monitoring data block by means of convolutional calculation;

[0061] The convolutional calculation formula is as follows:

[0062] f = S × ω + b;

[0063] Where, S represents the input cable historical monitoring data block, ω represents the weights of the corresponding convolutional kernel, b represents the bias value, and f represents the calculated data features.

[0064] Preferably, the summary drone collects cable monitoring data in real time and analyzes it through a trained security monitoring model. To determine whether there are abnormalities in the cable monitoring data collected in real time, the following steps are included:

[0065] Calculate the similarity between the data features obtained after analyzing the cable monitoring data collected in real time and the saved data features, and set a similarity threshold;

[0066] Compare the features extracted in real time with the saved data features through a similarity calculation formula;

[0067] The similarity calculation formula is as follows:

[0068]

[0069] where, represents the data feature and the data feature the similarity value between them, represents the data feature obtained after analyzing the cable monitoring data collected in real time, represents the saved data feature;

[0070] Based on the set similarity threshold, when the similarity between the data features obtained after analyzing the cable monitoring data collected in real time and the saved data features is less than the set similarity threshold, it indicates that there are abnormalities in the cable monitoring data collected in real time; otherwise, there are no abnormalities.

[0071] The present invention also discloses a cable safety monitoring system based on autonomous underground drone inspection, which is used to implement a cable safety monitoring method based on autonomous underground drone inspection. The system includes: a data collection module, a path planning module, a data analysis module, a drone scheduling module, and a safety monitoring module;

[0072] The data collection module is used to collect cable historical monitoring data in real time and cable monitoring data collected by the drone in real time;

[0073] The path planning module is used to construct a cable monitoring topology map and plan the monitoring path based on the constructed cable monitoring topology map;

[0074] The data analysis module is used to perform training and analysis on the collected cable historical monitoring data and cable monitoring data collected in real time;

[0075] The drone scheduling module is used to calculate the energy of the drone and match it with the planned path, and schedule the drone according to the matching result;

[0076] The safety monitoring module is used to perform safety monitoring and processing according to the analysis result of the cable monitoring data collected in real time.

[0077] The beneficial effects of the present invention are as follows:

[0078] (1) By setting the cable safety monitoring area, constructing a cable monitoring topology map based on the positions of each monitoring node within the set cable safety monitoring area, arranging for drones to collect cable monitoring data in real time based on the constructed cable monitoring topology map; collecting cable historical monitoring data in real time and constructing a sample set, training the cable historical monitoring data in the sample set to obtain a trained safety monitoring model, finally summarizing the cable monitoring data collected by drones in real time, and analyzing it through the trained safety monitoring model to determine whether there are abnormalities in the cable monitoring data collected in real time. After detecting an abnormality, arranging for staff to handle it, which improves the accuracy of drone inspection and monitoring.

[0079] (2) By setting the cable safety monitoring area and collecting the path data within the cable safety monitoring area in real time, setting the positions of monitoring points, and finally saving the path data through a data quantization method and constructing a cable monitoring topology map, the rationality of cable safety monitoring is improved.

[0080] (3) By planning each cable line in the constructed cable monitoring topology map through a path planning algorithm to obtain the shortest path between each monitoring node, and completing the summarization of the lines of the entire cable monitoring topology map through continuous iteration; after the summarization is completed, collecting the current drone status data in real time, scheduling the drones based on the obtained shortest path between each monitoring node, and after the scheduling is completed, arranging the corresponding drones to conduct inspections and collecting cable monitoring data in real time, which improves the effectiveness of drone inspections.

[0081] (4) By dividing the cable historical monitoring data collected in real time and inputting the divided cable historical monitoring data into a convolutional neural network, and analyzing the cable historical monitoring data through the convolutional neural network and outputting data features, the analysis efficiency of drone monitoring data is improved.

[0082] (5) By summarizing the cable monitoring data collected by drones in real time and analyzing it through the trained safety monitoring model, and comparing the features extracted in real time with the saved data features through a similarity calculation formula, the security of cable safety monitoring is improved. Description of the Drawings

[0083] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0084] Figure 1 It is a schematic flowchart of the cable safety monitoring method for autonomous inspection of underground drones of the present invention. Detailed implementation manners

[0085] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0086] In the specific embodiments of the present invention,

[0087] Referring to Figure 1 as shown, the present invention provides a cable safety monitoring method based on autonomous inspection of underground drones, including the following steps:

[0088] S1. Set the cable safety monitoring area, and construct a cable monitoring topology map based on the positions of each monitoring node in the set cable safety monitoring area;

[0089] S2. Based on the constructed cable monitoring topology map, arrange drones to collect cable monitoring data in real time;

[0090] S21. Plan each cable line in the constructed cable monitoring topology map through a path planning algorithm to obtain the shortest path between each monitoring node;

[0091] S22. Real-time collect the current drone status data, and schedule the drone based on the obtained shortest path between each monitoring node;

[0092] S23. After the scheduling is completed, arrange the corresponding drone for inspection and collect cable monitoring data in real time;

[0093] S3. Real-time collect cable historical monitoring data and construct a sample set, and train the cable historical monitoring data in the sample set to obtain a trained safety monitoring model;

[0094] S4. Summarize the cable monitoring data collected by the drone in real time, and analyze it through the trained safety monitoring model to determine whether there is an abnormality in the cable monitoring data collected in real time;

[0095] S5. After detecting an anomaly, arrange for staff to handle it.

[0096] Further, with reference to Figure 1 As shown, set the cable safety monitoring area, and based on the positions of each monitoring node within the set cable safety monitoring area, construct a cable monitoring topology map, including the following steps:

[0097] Set the cable safety monitoring area and collect the path data within the cable safety monitoring area in real time;

[0098] Set the intersection points of the paths within the cable safety monitoring area as monitoring nodes;

[0099] Set the set of path data within the cable safety monitoring area as: e = {e1, e2,..., e m};

[0100] where e represents the set of path data within the cable safety monitoring area, and e m represents the m-th path data within the cable safety monitoring area;

[0101] Further, set the path data e m to be composed of a triple {a m , z m , d m}, where a m represents the starting monitoring point of the path data, z m represents the ending monitoring point of the path data, and d m represents the distance of the path;

[0102] Further, construct a cable monitoring topology map based on the collected set of path data;

[0103] Further, with reference to Figure 1 As shown, plan each cable line in the constructed cable monitoring topology map through a path planning algorithm to obtain the shortest paths between each monitoring node, including the following steps:

[0104] S211. Initialize the path matrix based on the constructed cable monitoring topology map, and set the data within the matrix as the distance of the path;

[0105] S212. Randomly select a monitoring point, and based on the selected monitoring point as the starting point, update the path matrix again. Set that if there is a direct path between the monitoring point and the starting point, then set the distance of the direct path as the corresponding updated matrix data;

[0106] S213. Select the path with the smallest matrix data, take the monitoring points on this path as a whole, update the path matrix again until all monitoring points are traversed, and output the shortest paths between each monitoring node;

[0107] S214. Save the shortest paths between the monitored nodes of the output, and remove the corresponding paths in the constructed cable monitoring topology map. If there is only one path between a monitoring point and other monitoring points, the corresponding monitoring points in the cable monitoring topology map shall also be removed. After removing the corresponding paths, iteratively execute steps S212 - S213 until there is no path between the monitored nodes, and output the saved multiple groups of shortest paths between the monitored nodes;

[0108] Further, referring to Figure 1 as shown, real - time collect the current UAV status data, and schedule the UAV based on the shortest paths between the monitored nodes obtained, including the following steps:

[0109] S221. Real - time collect the UAV status data and calculate the inspection cost of the UAV;

[0110] Real - time collecting the UAV status data includes: the battery power of the UAV and the working status of the UAV;

[0111] Setting the inspection cost of the UAV includes: the battery power of the UAV and the inspection time of the UAV;

[0112] Among them, the UAV inspection time = data feedback time + data processing time + network delay time;

[0113] The battery power of the UAV = flight power consumption + data processing power consumption + data feedback power consumption. The formula for the inspection cost of the UAV is as follows:

[0114]

[0115] where, Z j represents the inspection cost of the j - th UAV, E j represents the power required for the j - th UAV to perform an inspection, T j represents the inspection time using the j - th UAV, represents the energy consumption weight of the j - th UAV, represents the time weight of the j - th UAV;

[0116] S222. Schedule the UAV based on the calculated inspection cost of the UAV and the shortest paths between multiple groups of monitored nodes;

[0117] S2221. Initialize the algorithm parameters and perform chromosome encoding on the shortest paths between multiple groups of monitored nodes;

[0118] S2222. Construct the initial population by enabling the saturated transportation mode;

[0119] The saturation transportation method is that a random drone is arranged for inspection along the shortest path between each group of monitoring nodes;

[0120] S2223. Calculate the population fitness through the fitness function;

[0121] Set each individual to represent the shortest path between each group of monitoring nodes, and the shortest path between each group of monitoring nodes represents the path of the drone after inspecting all monitoring points in the current inspection path;

[0122] Set the fitness function:

[0123] Among them, Fit(x) represents the fitness of individual x, and Z represents the inspection cost of the drone;

[0124] S2224. Genetic operator selection operation;

[0125] Set the probability of an individual being selected as p(x i );

[0126]

[0127] Among them, I is the population size, and Fit(x i ) is the fitness of the i-th individual;

[0128] S2225. Output the optimal solution of the problem;

[0129] Set the maximum number of iterations Y. Set that when the chromosome encoding converges after Y iterations, terminate the hybrid genetic algorithm, and use the chromosome encoding output at the last time as the inspection path for scheduling the drone;

[0130] Furthermore, as shown in Figure 1 , the following steps are included to collect the cable historical monitoring data in real time and construct a sample set, and train the cable historical monitoring data in the sample set to obtain a trained safety monitoring model:

[0131] S31. Collect the cable historical monitoring data, and summarize and evenly divide the collected cable historical monitoring data into two groups, one group as the training group and one group as the verification group;

[0132] S32. Divide each cable historical monitoring data in the training group into n cable historical monitoring data blocks by an equal-proportion division method;

[0133] S33. Input the divided cable historical monitoring data blocks into the convolutional neural network and train through the convolutional neural network;

[0134] After the convolutional neural network receives the cable historical monitoring data block, set the convolution kernel size and initial weights of the convolutional layer in the convolutional neural network, and calculate the data features corresponding to the cable historical monitoring data block by means of convolutional calculation;

[0135] The convolution calculation formula is as follows:

[0136] f = S × ω + b;

[0137] Among them, S represents the input cable historical monitoring data block, ω represents the weights of the corresponding convolution kernel, b represents the bias value, and f represents the calculated data features;

[0138] S34. Expand and combine the data features through the fully connected layer to obtain a feature array and save it;

[0139] S35. Input the cable historical monitoring data in the verification group into the convolutional neural network, and match the output result of the verification group with the output result of the training group;

[0140] Set the matching success threshold. When the matching result is greater than or equal to the set threshold, output the current convolutional neural network as the recognition model. When the matching result is less than the set threshold, adjust the weights of the convolutional neural network until the matching result is greater than the set threshold;

[0141] Furthermore, set the adjusted convolutional neural network as the trained safety monitoring model;

[0142] Furthermore, refer to Figure 1 As shown, summarize the cable monitoring data collected by the drone in real time, and analyze it through the trained safety monitoring model to determine whether there are abnormalities in the cable monitoring data collected in real time, including the following steps:

[0143] Calculate the similarity between the data features after analyzing the cable monitoring data collected in real time and the saved data features, and set the similarity threshold;

[0144] Compare the features extracted in real time with the saved data features through the similarity calculation formula;

[0145] The similarity calculation formula is as follows:

[0146]

[0147] Among them, represents the similarity value between the data features and the data features ; represents the data features after analyzing the cable monitoring data collected in real time, represents the saved data features;

[0148] Furthermore, based on the set similarity threshold, when the similarity between the data features of the cable monitoring data collected in real time after analysis and the saved data features is less than the set similarity threshold, it indicates that there are abnormalities in the cable monitoring data collected in real time; otherwise, there are no abnormalities.

[0149] In a specific embodiment, the cable safety monitoring system based on autonomous underground drone inspection is used to implement a cable safety monitoring method based on autonomous underground drone inspection. The system includes: a data collection module, a path planning module, a data analysis module, a drone scheduling module, and a safety monitoring module.

[0150] The data collection module is used to collect cable historical monitoring data in real time and cable monitoring data collected by drones in real time.

[0151] The path planning module is used to construct a cable monitoring topology map and plan the monitoring path based on the constructed cable monitoring topology map.

[0152] The data analysis module is used to perform training analysis on the collected cable historical monitoring data and cable monitoring data collected in real time.

[0153] The drone scheduling module is used to calculate the energy of the drone and match it with the planned path, and schedule the drone according to the matching result.

[0154] The safety monitoring module is used to perform safety monitoring and processing according to the analysis results of the cable monitoring data collected in real time.

[0155] It should be noted that

[0156] The above content is only an example and explanation of the concept of the present invention. Those skilled in the art of this technology can make various modifications or supplements to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A cable safety monitoring method based on autonomous inspection by underground drones, characterized in that, Including the following steps: S1. Set the cable safety monitoring area, and based on the positions of each monitoring node within the set cable safety monitoring area, construct a cable monitoring topology map; S2. Based on the constructed cable monitoring topology map, arrange for drones to collect cable monitoring data in real time; S21. Through a path planning algorithm, plan each cable line in the constructed cable monitoring topology map to obtain the shortest path between each monitoring node; S22. Real-time collect the current drone status data, and schedule the drones based on the obtained shortest path between each monitoring node; S23. After the scheduling is completed, arrange the corresponding drones for inspection and collect cable monitoring data in real time; S3. Real-time collect cable historical monitoring data and construct a sample set, and train the cable historical monitoring data in the sample set to obtain a trained safety monitoring model; S4. Summarize the cable monitoring data collected by the drones in real time, and analyze it through the trained safety monitoring model to determine whether there are any abnormalities in the cable monitoring data collected in real time; S5. After detecting an abnormality, arrange for staff to handle it.

2. The cable safety monitoring method based on autonomous inspection of underground drones according to claim 1, wherein, The step of setting the cable safety monitoring area and constructing a cable monitoring topology map based on the positions of each monitoring node within the set cable safety monitoring area includes the following steps: Set the cable safety monitoring area and real-time collect the path data within the cable safety monitoring area; Set the intersection points of the paths within the cable safety monitoring area as monitoring nodes; Set the path data set within the cable safety monitoring area as: e = {e1, e2,..., e m}; Among them, e represents the set of path data within the cable safety monitoring area, and e m represents the m-th path data within the cable safety monitoring area; Set path data e m Consists of a triple {a m , z m , d m}, where a m represents the starting monitoring point of the path data, z m represents the ending monitoring point of the path data, and d m represents the distance of the path; Based on the collected path data set, construct a cable monitoring topology map.

3. A cable safety monitoring method based on autonomous inspection of underground drones according to claim 1, characterized in that, The step of planning each cable line in the constructed cable monitoring topology map through a path planning algorithm to obtain the shortest path between each monitoring node includes the following steps: S211. Initialize the path matrix based on the constructed cable monitoring topology map, and set the data in the matrix as the distance of the path; S212. Randomly select a monitoring point, based on the selected monitoring point as the starting point, re-update the path matrix, and set that if there is a direct path between the monitoring point and the starting point, then set the distance of the direct path as the corresponding updated matrix data; S213. Select the path with the smallest matrix data, take the monitoring points on this path as a whole, re-update the path matrix until all monitoring points are traversed, and output the shortest path between each monitoring node; S214. Save the output shortest path between each monitoring node, and remove the corresponding path in the constructed cable monitoring topology map. If there is only one path between a monitoring point and other monitoring points, then the corresponding monitoring point in the cable monitoring topology map is also removed. After removing the corresponding path, iteratively execute steps S212 - S213 until there is no path between each monitoring node, and output the saved multiple groups of shortest paths between each monitoring node.

4. A cable safety monitoring method based on autonomous inspection of underground drones according to claim 1, characterized in that, The step of real-time collecting the current drone status data and scheduling the drones based on the obtained shortest path between each monitoring node includes the following steps: S221. Real-time collect the drone status data and calculate the inspection cost of the drones; S222. Schedule the drones based on the calculated inspection cost of the drones and the multiple groups of shortest paths between each monitoring node.

5. The cable safety monitoring method based on autonomous inspection of underground drones according to claim 4, characterized in that, The steps of collecting the state data of the UAV in real time and calculating the inspection cost of the UAV are as follows: Collecting the state data of the UAV in real time includes: the battery power of the UAV and the working state of the UAV; Setting the inspection cost of the UAV includes: the battery power of the UAV and the inspection time of the UAV; Among them, the UAV inspection time = data feedback time + data processing time + network delay time; The UAV battery power = flight power consumption + data processing power consumption + data feedback power consumption The formula for the inspection cost of the UAV is as follows: Among them, Z j represents the inspection cost of the j-th UAV, E j represents the power required for the j-th UAV inspection, T j represents the inspection time using the j-th UAV, represents the energy consumption weight of the j-th UAV, represents the time weight of the j-th UAV.

6. The cable safety monitoring method based on autonomous inspection by underground drones according to claim 4, characterized in that, The steps of scheduling the UAV based on the calculated inspection cost of the UAV and the shortest paths between multiple groups of monitoring nodes are as follows: S222. Schedule the UAV based on the calculated inspection cost of the UAV and the shortest paths between multiple groups of monitoring nodes; S2221. Initialize the algorithm parameters and perform chromosome coding on the shortest paths between multiple groups of monitoring nodes; S2222. Construct an initial population by enabling the saturated transportation mode; The saturated transportation mode is to arrange a random UAV for inspection on the shortest path between each group of monitoring nodes; S2223. Calculate the population fitness through the fitness function; Set each individual to represent the shortest path between a group of monitoring nodes, and the shortest path between each group of monitoring nodes represents the path of the UAV after inspecting all monitoring points in the current inspection path; Set the fitness function: Among them, Fit(x) represents the fitness of individual x, and Z represents the inspection cost of the UAV; S2224. Genetic operator selection operation; Set the probability that a monomer is selected to be p(x i ); Among them, I is the population size, and Fit(x i ) is the fitness of the i-th monomer; S2225. Output the optimal solution of the problem; Set the maximum number of iterations Y, and set that when the chromosome coding converges after Y iterations, terminate the hybrid genetic algorithm, and use the chromosome coding output last time as the inspection path for scheduling the UAV.

7. A cable safety monitoring method based on autonomous inspection by an underground drone according to claim 1, characterized in that, The steps of collecting the historical monitoring data of the cable in real time, constructing a sample set, and training the historical monitoring data of the cable in the sample set to obtain a trained safety monitoring model are as follows: S31. Collect the historical monitoring data of the cable, summarize the collected historical monitoring data of the cable and divide it evenly into two groups, one group as the training group and one group as the verification group; S32. Divide each historical monitoring data of the cable in the training group into n historical monitoring data blocks by the equal proportion division method; S33. Input the divided historical monitoring data blocks of the cable into the convolutional neural network and train through the convolutional neural network; S35. Input the historical monitoring data of the cable in the verification group into the convolutional neural network, and match the output result of the verification group with the output result of the training group; Set the matching success threshold. When the matching result is greater than or equal to the set threshold, output the current convolutional neural network as the recognition model. When the matching result is less than the set threshold, adjust the weights of the convolutional neural network until the matching result is greater than the set threshold; Set the adjusted convolutional neural network as the trained safety monitoring model.

8. A cable safety monitoring method based on autonomous inspection by an underground drone according to claim 1, characterized in that The steps of inputting the divided historical monitoring data blocks of the cable into the convolutional neural network and training through the convolutional neural network are as follows: After the convolutional neural network receives the cable historical monitoring data block, set the convolution kernel size and initial weights of the convolutional layer in the convolutional neural network, and calculate the data features corresponding to the cable historical monitoring data block by means of convolution calculation; The convolution calculation formula is as follows: f = S × ω + b; Among them, S represents the input cable historical monitoring data block, ω represents the weights of the corresponding convolution kernel, b represents the bias value, and f represents the calculated data features.

9. A cable safety monitoring method based on autonomous inspection of underground drones according to claim 1, characterized in that, The summarized drone collects cable monitoring data in real time and analyzes it through the trained safety monitoring model. Judging whether there are abnormalities in the cable monitoring data collected in real time includes the following steps: Calculate the similarity between the data features after analyzing the cable monitoring data collected in real time and the saved data features, and set a similarity threshold; Compare the features extracted in real time with the saved data features through the similarity calculation formula; The similarity calculation formula is as follows: Among them, represents the similarity value between data features and data features represents the data features after analyzing the cable monitoring data collected in real time, represents the saved data features; Based on the set similarity threshold, when the similarity between the data features after analyzing the cable monitoring data collected in real time and the saved data features is less than the set similarity threshold, it means that there are abnormalities in the cable monitoring data collected in real time, otherwise there are no abnormalities.

10. A system for implementing the cable safety monitoring method based on autonomous inspection by an underground drone according to any one of claims 1-9, characterized in that, Including: Data collection module, path planning module, data analysis module, drone scheduling module and safety monitoring module; The data collection module is used to collect cable historical monitoring data in real time and cable monitoring data collected by drones in real time; The path planning module is used to construct a cable monitoring topology map and plan the monitoring path based on the constructed cable monitoring topology map; The data analysis module is used to train and analyze the collected cable historical monitoring data and cable monitoring data collected in real time; The drone scheduling module is used to calculate the energy of the drone and match it with the planned path, and schedule the drone according to the matching result; The safety monitoring module is used to perform safety monitoring and processing according to the analysis results of the cable monitoring data collected in real time.

Images