Self-adaptive patrol inspection method and unmanned aerial vehicle

Abstract

The invention discloses a self-adaptive patrol inspection method and an unmanned aerial vehicle, and relates to the technical field of unmanned aerial vehicle patrol inspection.The method comprises the steps that a high-precision live-action three-dimensional map is modified in real time; establishing a calculation mode of the high-risk zone; calculating to obtain a high-risk index of the high-risk zone, and setting patrol frequency of the high-risk zone; the AI technology is used for analyzing the abnormity type of the abnormity early warning, and an inspection scheme is formed for the abnormity early warning; establishing a task severity evaluation mode, obtaining the severity of abnormal early warning, and setting the execution sequence of the inspection schemes; deploying the non-patrol unmanned aerial vehicle which returns and the non-patrol unmanned aerial vehicle of the unmanned aerial vehicle, and executing the inspection scheme by the deployed non-patrol unmanned aerial vehicle. According to the method, different inspection schemes are formed for inspection of different severity degrees by estimating the inspection condition, it is guaranteed that more emergency anomalies are subjected to priority inspection, daily inspection is carried out on high-risk zones, and the comprehensiveness of inspection is further guaranteed.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) inspection technology, specifically to an adaptive patrol and inspection method and an unmanned aerial vehicle vehicle. Background Technology

[0002] Power line inspection utilizes computer vision and artificial intelligence technologies to replace or assist traditional manual inspection methods. It enables automatic identification, real-time early warning, precise location, and closed-loop management of power line operating status and surrounding environmental safety hazards, thereby improving the safety, efficiency, and intelligence of power grid operation and maintenance. Traditional inspection methods, limited by terrain and environment, often cannot quickly locate power anomalies, leading to prolonged inspection times. Using unmanned aerial vehicles (UAVs) to conduct inspections can significantly enhance efficiency.

[0003] However, when the inspection covers a large area and there are many locations that need to be inspected, the existing technology is insufficient to predict the inspection situation, making it difficult to allocate drones reasonably. This can easily lead to delays in inspections with more critical situations. At the same time, it can also easily result in insufficient drones being used for inspections, thus compromising the efficiency of the inspection. Summary of the Invention

[0004] To address the aforementioned technical problems, an adaptive patrol and inspection method and an unmanned vehicle are provided. This technical solution resolves the issues raised in the background section.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An adaptive patrol and inspection method includes: The drone vehicle is dustproof and off-road capable. It carries at least one drone, which communicates via satellite. The audio and video collected by the drone during inspections are transmitted back to the command center in real time. The number of drones in the unmanned vehicle is obtained, and based on historical data, the drones of the unmanned vehicle are divided into patrol drones and non-patrol drones; Based on the information collected by the UAV data acquisition system, the high-precision real-scene 3D map is modified in real time. Based on historical data of anomalies and high-precision real-scene 3D maps, a calculation model for high-risk areas is established. Obtain the actual data of the high-risk area required in the calculation mode of the high-risk area, substitute the actual data of the high-risk area into the calculation mode of the high-risk area, calculate the high-risk index of the high-risk area, set the patrol frequency of the high-risk area based on the high-risk index of the high-risk area, and patrol drones patrol the high-risk area. When the unmanned vehicle receives an anomaly warning, it uses AI technology to analyze the anomaly category of the warning and formulate an inspection plan accordingly. Establish a task severity assessment model, obtain the severity of anomaly warnings based on the task severity assessment model, and set the execution order of inspection plans based on the severity of the anomaly warnings corresponding to the inspection plans. The inspection plan is executed according to the order of execution. Based on the inspection plan, non-patrol drones returning to base and non-patrol drones of unmanned vehicles are deployed, and the deployed non-patrol drones execute the inspection plan.

[0006] Preferably, the real-time modification of the high-precision real-scene 3D map based on the information collected by the UAV data acquisition system includes the following steps: At least one image acquired by the UAV data acquisition system is used as a feature image; The acquisition location of the feature image is obtained. The image area in the high-precision real-scene 3D map that matches the acquisition location is used as the image to be compared. The feature image corresponding to the same acquisition location is paired with the image to be compared. The degree of change is calculated using the formula for the degree of change, and the rate of change is calculated using the formula for the rate of change. Replace the corresponding comparison image in the high-precision real-scene 3D map with the feature image whose degree of change is higher than the first preset value or whose rate of change is higher than the second preset value. The first and second preset values ​​are set based on empirical data. The formula for the degree of change is as follows:

[0007] In the formula, C represents the degree of change, and I represents the number of pixels that differ between the image to be compared and the corresponding feature image. The formula for the rate of change is as follows:

[0008] In the formula, V is the rate of change. The value is the median, and i is the index. Let represent the degree of change of the i-th feature image, and n be the total number of feature images.

[0009] Preferably, the calculation model for establishing high-risk areas includes the following steps: Areas where anomalies occur more than the preset number of times are designated as high-risk zones. The number of anomalies occurring in high-risk areas is used as a risk reference factor; Calculate the potential hazard factor of a high-risk area using the potential hazard factor formula: Calculate the site hazard coefficient using the site hazard coefficient formula: Using the analytic hierarchy process (AHP), the weights of the hazard reference coefficient, the site hazard coefficient, and the potential hazard coefficient are obtained. The high-risk index of the high-risk area is then calculated using the high-risk index calculation formula. The formula for calculating the high-risk index is as follows:

[0010] In the formula, Q is the high-risk index, t is the risk reference coefficient, o is the site risk coefficient, r is the potential risk coefficient, T is the weight of the risk reference coefficient, O is the weight of the site risk coefficient, and R is the weight of the potential risk coefficient. The formula for the potential hazard factor is as follows:

[0011] In the formula, r is the potential hazard coefficient, p is the power line density in the high-risk area, f is the dead tree density in the high-risk area, and e is the natural constant. The formula for the site risk factor is as follows:

[0012] In the formula, o is the site hazard coefficient, w is the vegetation cover density of the high-risk area, and s is the area in the high-risk area where the power line density is lower than the preset density.

[0013] Preferably, setting the patrol frequency of high-risk areas based on the high-risk index of high-risk areas includes the following steps: The high-risk index of high-risk areas is rounded down to the nearest whole number and recorded as the whole index. The patrol frequency of high-risk areas is also recorded as the whole index. Based on the number of high-risk areas and the number of drones in the drone vehicle, the maximum patrol frequency is calculated using the formula for calculating the maximum patrol frequency. The rate of change of feature images containing high-risk areas is obtained as the feature change rate. When the rate of change of characteristics exceeds the preset rate, the patrol frequency of high-risk areas will be adjusted to the highest patrol frequency. The formula for calculating the maximum patrol frequency is as follows:

[0014] In the formula, H represents the highest patrol frequency, N represents the number of high-risk areas, and L represents the number of patrol drones in the drone vehicle.

[0015] Preferably, the step of using AI technology to analyze the anomaly category of the anomaly warning and formulate an inspection plan for the anomaly warning includes the following steps: Based on the number of abnormal power lines, the degree of abnormality, and the area covered by the abnormality, the anomalies in the historical anomaly warnings are classified to obtain at least one anomaly category. Extract at least one sample anomaly from the anomaly category. Based on historical data, obtain historical anomalies that are consistent with the sample anomalies. Use the union of the solutions to the historical anomalies as the inspection plan for the anomaly category. AI technology is used to analyze the anomaly categories of anomaly warnings, and the inspection plans corresponding to the anomaly categories of anomaly warnings are used as the inspection plans for anomaly warnings.

[0016] Preferably, the establishment of the task severity assessment model includes the following steps: Based on the acquired task information, AI is used to analyze the number of abnormal power lines and the number of potential abnormal power lines, and the sum of the number of abnormal power lines and the number of potential abnormal power lines is recorded as the line hazard level. Based on the acquired task information, AI analysis is used to determine the extent of line damage and the geographical area where the line damage occurred. Using the analytic hierarchy process (AHP), we obtained the weights of the line hazard level, the line damage degree, and the geographical area of ​​the line damage. The severity of anomaly warnings is calculated based on the formula for calculating the severity of anomaly warnings. The formula for calculating severity is as follows:

[0017] In the formula, s represents the severity, d represents the line hazard level, b represents the degree of line damage, m represents the geographical area where the line damage occurred, D represents the weight of the line hazard level, B represents the weight of the degree of line damage, and M represents the weight of the geographical area where the line damage occurred.

[0018] Preferably, the process of determining the severity of anomaly alerts based on the task severity assessment model includes the following steps: By substituting the actual data of the tasks for which anomaly warnings are required in the task severity assessment model into the model, the severity of the anomaly warnings can be obtained.

[0019] Preferably, setting the execution order of the inspection plan based on the severity of the abnormal warning corresponding to the inspection plan includes the following steps: Compare the severity of anomaly warnings and prioritize tasks with higher severity warnings. If the severity of the abnormal warnings for two tasks is the same, then compare the line hazard of the two tasks and prioritize the task with the higher line hazard. If the severity of the abnormal warning and the line hazard level of the two tasks are the same, then compare the area of ​​the line damage that occurred in the two tasks, and prioritize the task with the higher area of ​​line damage. If the severity of the abnormal warning, the degree of line danger, and the area of ​​line damage are the same for two tasks, then the task with the higher degree of line damage will be prioritized.

[0020] Preferably, the deployment of returning drones and drone vehicles based on the inspection plan includes the following steps: Using the minimum number of drones calculation formula, calculate the number of drones required to execute the inspection plan, and use this as the number of features; Obtain the distance from the location of the inspection plan to the unmanned vehicle as the feature distance; Non-patrol drones that return to their destination at a distance less than the characteristic distance from the location of the inspection plan are designated as target drones, and the target drones are deployed to execute the inspection plan. The number of target drones is counted as the target count. The number of compensations is obtained by subtracting the target count from the number of features. Non-patrol drones in the unmanned vehicle are deployed to perform inspection plans, and the number of non-patrol drones in the deployed unmanned vehicle is equal to the number of compensation drones. The formula for calculating the minimum number of drones is as follows:

[0021] In the formula, I represents the number of features, S represents the severity of the abnormal warning corresponding to the inspection plan, and Z represents the number of non-patrol drones in the unmanned vehicle.

[0022] An adaptive patrol and inspection drone vehicle performs patrol and inspection operations according to the aforementioned adaptive patrol and inspection method.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: By forecasting the inspection situation, different inspection plans are formulated for inspections of varying severity. The execution order of the inspection plans is set according to the severity of the corresponding anomaly warnings. This ensures that the number of drones matched with the anomalies is appropriate, and that more urgent anomalies are prioritized for inspection. In addition, drones returning to base and drones in drone vehicles are deployed simultaneously, avoiding situations where only drones in drone vehicles are deployed. This reduces the overall flight distance of drones to some extent. At the same time, daily patrols are conducted in high-risk areas to further ensure the comprehensiveness of the inspections. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the adaptive patrol and inspection method of the present invention. Figure 2 This is a schematic diagram illustrating the process of real-time modification of a high-precision real-scene 3D map based on information collected by the UAV data acquisition system according to the present invention. Figure 3 This is a schematic diagram of the process for establishing a calculation model for high-risk areas according to the present invention; Figure 4 This is a flowchart illustrating the process of setting the patrol frequency of high-risk areas based on the high-risk index of high-risk areas according to the present invention. Figure 5 This is a flowchart illustrating the process of using AI technology to analyze the anomaly categories of anomaly warnings and to generate inspection plans based on anomaly warnings, as described in this invention. Figure 6 This is a schematic diagram of the process for establishing a task severity assessment model according to the present invention; Figure 7 This is a flowchart illustrating how the execution order of an inspection plan is set based on the severity of the abnormality warning corresponding to the inspection plan in this invention. Figure 8 This is a schematic diagram illustrating the process of deploying returning drones and drones in drone vehicles based on the inspection scheme of the present invention. Figure 9 This is a side view structural diagram of the adaptive patrol and inspection unmanned vehicle of the present invention; Figure 10 This is a front view structural diagram of the adaptive patrol and inspection drone vehicle of the present invention; Figure 11 This is a rear-view structural diagram of the adaptive patrol and inspection unmanned vehicle of the present invention; Figure 12 This is a top view schematic diagram of the drone-mounted adaptive patrol and inspection drone vehicle of the present invention. Figure 13 This is a schematic diagram of the structure of the drone mounted on the adaptive patrol and inspection drone vehicle of the present invention. Detailed Implementation

[0025] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0026] Reference Figure 1 As shown, an adaptive patrol and inspection method includes: The drone vehicle is dustproof and off-road capable. It carries at least one drone, which communicates via satellite. The audio and video collected by the drone during inspections are transmitted back to the command center in real time. The number of drones in the unmanned vehicle is obtained, and based on historical data, the drones of the unmanned vehicle are divided into patrol drones and non-patrol drones; Based on the information collected by the UAV data acquisition system, the high-precision real-scene 3D map is modified in real time. Based on historical data of anomalies and high-precision real-scene 3D maps, a calculation model for high-risk areas is established. Obtain the actual data of the high-risk area required in the calculation mode of the high-risk area, substitute the actual data of the high-risk area into the calculation mode of the high-risk area, calculate the high-risk index of the high-risk area, set the patrol frequency of the high-risk area based on the high-risk index of the high-risk area, and patrol drones patrol the high-risk area. When the unmanned vehicle receives an anomaly warning, it uses AI technology to analyze the anomaly category of the warning and formulate an inspection plan accordingly. Establish a task severity assessment model, obtain the severity of anomaly warnings based on the task severity assessment model, and set the execution order of inspection plans based on the severity of the anomaly warnings corresponding to the inspection plans. The inspection plan is executed according to the order of execution. Based on the inspection plan, non-patrol drones returning to base and non-patrol drones of unmanned vehicles are deployed, and the deployed non-patrol drones execute the inspection plan.

[0027] When drones on unmanned aerial vehicles (UAVs) are conducting inspections, they cover a large area. Therefore, multiple inspection warnings may appear in parallel or at short intervals. Since the number of drones on an unmanned aerial vehicle is not unlimited, and the circumstances of abnormal warnings are also different, it is necessary to estimate the abnormal warnings and match the corresponding drones. Otherwise, it will lead to an imbalance in drone allocation and a mismatch between the drones and the scale of the abnormal warnings. As a result, the deployed drones will not be able to complete the inspection of the abnormal warning locations in a timely manner. This is because the anomalies at the abnormal warning locations may be quite serious and large in scale. The time required for an insufficient number of drones to complete a comprehensive inspection will inevitably be longer. In cases where there are injured persons, a longer inspection time is not the best choice. At the same time, for high-risk areas, inspections should not be carried out only when problems occur. In order to avoid potential risks, regular patrols are required to deal with problems in their initial stages. This solution addresses all the aforementioned issues in a targeted manner.

[0028] Reference Figure 2 As shown, the real-time modification of a high-precision real-scene 3D map based on information collected by the UAV data acquisition system includes the following steps: At least one image acquired by the UAV data acquisition system is used as a feature image; The acquisition location of the feature image is obtained. The image area in the high-precision real-scene 3D map that matches the acquisition location is used as the image to be compared. The feature image corresponding to the same acquisition location is paired with the image to be compared. The degree of change is calculated using the formula for the degree of change, and the rate of change is calculated using the formula for the rate of change. The formula for the degree of change is as follows:

[0029] In the formula, C represents the degree of change, and I represents the number of pixels that differ between the image to be compared and the corresponding feature image. The formula for the rate of change is as follows:

[0030] In the formula, V is the rate of change. The value is the median, and i is the index. The degree of change of the i-th feature image is given by n, where n is the total number of feature images. Replace the corresponding comparison image in the high-precision real-scene 3D map with the feature image whose degree of change is higher than the first preset value or whose rate of change is higher than the second preset value. The first and second preset values ​​are set based on empirical data.

[0031] Since maps can change, but modifications do not require global changes, targeted modifications can be made at corresponding locations by comparing feature images with images to be compared. The basis for modification is the degree or rate of change. When one of these two factors is greater, the original high-precision real-scene 3D map image at the corresponding location becomes insufficient for reference and needs to be updated.

[0032] Reference Figure 3 As shown, establishing a calculation model for high-risk areas includes the following steps: Areas where anomalies occur more than the preset number of times are designated as high-risk zones. The number of anomalies occurring in high-risk areas is used as a risk reference factor; Calculate the potential hazard factor of a high-risk area using the potential hazard factor formula: Calculate the site hazard coefficient using the site hazard coefficient formula: Using the analytic hierarchy process (AHP), the weights of the hazard reference coefficient, the site hazard coefficient, and the potential hazard coefficient are obtained. The high-risk index of the high-risk area is then calculated using the high-risk index calculation formula. The formula for calculating the high-risk index is as follows:

[0033] In the formula, Q is the high-risk index, t is the risk reference coefficient, o is the site risk coefficient, r is the potential risk coefficient, T is the weight of the risk reference coefficient, O is the weight of the site risk coefficient, and R is the weight of the potential risk coefficient. The formula for the potential hazard factor is as follows:

[0034] In the formula, r is the potential hazard coefficient, p is the power line density in the high-risk area, f is the dead tree density in the high-risk area, and e is the natural constant. The formula for the site risk factor is as follows:

[0035] In the formula, o is the site hazard coefficient, w is the vegetation cover density of the high-risk area, and s is the area in the high-risk area where the power line density is lower than the preset density.

[0036] The high-risk zone calculation mode is used to set the patrol frequency for high-risk zones. Although they are all high-risk zones, their levels of danger are different, and therefore, the probability of anomalies is also different. Based on this, the patrol frequency of more dangerous high-risk zones needs to be increased, while the patrol frequency of less dangerous high-risk zones needs to be decreased. However, due to the limitation on the number of drones in the unmanned vehicle, the patrol frequency of high-risk zones cannot be increased indefinitely. Therefore, the maximum patrol frequency is calculated in subsequent steps. When the characteristic change rate is very high, it indicates that there are significant changes in the high-risk area, and therefore, anomalies are likely to occur. Therefore, due to the limitation on the number of drones, the maximum patrol frequency is used for patrolling.

[0037] Reference Figure 4 As shown, setting the patrol frequency for high-risk areas based on their high-risk index includes the following steps: The high-risk index of high-risk areas is rounded down to the nearest whole number and recorded as the whole index. The patrol frequency of high-risk areas is also recorded as the whole index. Based on the number of high-risk areas and the number of drones in the drone vehicle, the maximum patrol frequency is calculated using the formula for calculating the maximum patrol frequency. The rate of change of feature images containing high-risk areas is obtained as the feature change rate. When the rate of change of characteristics exceeds the preset rate, the patrol frequency of high-risk areas will be adjusted to the highest patrol frequency. The formula for calculating the maximum patrol frequency is as follows:

[0038] In the formula, H represents the highest patrol frequency, N represents the number of high-risk areas, and L represents the number of patrol drones in the drone vehicle.

[0039] Reference Figure 5 As shown, using AI technology to analyze the anomaly categories of anomaly warnings and formulate inspection plans for these warnings includes the following steps: Based on the number of abnormal power lines, the degree of abnormality, and the area covered by the abnormality, the anomalies in the historical anomaly warnings are classified to obtain at least one anomaly category. Extract at least one sample anomaly from the anomaly category. Based on historical data, obtain historical anomalies that are consistent with the sample anomalies. Use the union of the solutions to the historical anomalies as the inspection plan for the anomaly category. AI technology is used to analyze the anomaly categories of anomaly warnings, and the inspection plans corresponding to the anomaly categories of anomaly warnings are used as the inspection plans for anomaly warnings.

[0040] Different abnormal warnings require different inspection plans. The same inspection plan cannot be used for all abnormal warnings, as this will lead to insufficient matching of inspections, which may delay more serious inspections and cause a series of adverse consequences.

[0041] Reference Figure 6 As shown, establishing a task severity assessment model includes the following steps: Based on the acquired task information, AI is used to analyze the number of abnormal power lines and the number of potential abnormal power lines, and the sum of the number of abnormal power lines and the number of potential abnormal power lines is recorded as the line hazard level. Based on the acquired task information, AI analysis is used to determine the extent of line damage and the geographical area where the line damage occurred. Using the analytic hierarchy process (AHP), we obtained the weights of the line hazard level, the line damage degree, and the geographical area of ​​the line damage. The severity of anomaly warnings is calculated based on the formula for calculating the severity of anomaly warnings. The formula for calculating severity is as follows:

[0042] In the formula, s represents the severity, d represents the line hazard level, b represents the degree of line damage, m represents the geographical area where the line damage occurred, D represents the weight of the line hazard level, B represents the weight of the degree of line damage, and M represents the weight of the geographical area where the line damage occurred.

[0043] Since the tasks sent to the unmanned vehicle during task execution are not just one, but may be multiple and concurrent, and many more tasks follow, the task execution order must be set due to the limited number of drones in the unmanned vehicle. Otherwise, more serious tasks may not be executed, which would have more serious consequences and greater losses. Therefore, it is necessary to generate a corresponding task severity assessment mode to obtain the severity of abnormal warnings, and then set the execution order of the inspection plan based on this.

[0044] Based on the task severity assessment model, the severity of the anomaly warning is determined through the following steps: By substituting the actual data of the tasks for which anomaly warnings are required in the task severity assessment model into the model, the severity of the anomaly warnings can be obtained.

[0045] Reference Figure 7 As shown, based on the severity of the anomaly warning corresponding to the inspection plan, the execution sequence of the inspection plan includes the following steps: Compare the severity of anomaly warnings and prioritize tasks with higher severity warnings. If the severity of the abnormal warnings for two tasks is the same, then compare the line hazard of the two tasks and prioritize the task with the higher line hazard. If the severity of the abnormal warning and the line hazard level of the two tasks are the same, then compare the area of ​​the line damage that occurred in the two tasks, and prioritize the task with the higher area of ​​line damage. If the severity of the abnormal warning, the degree of line danger, and the area of ​​line damage are the same for two tasks, then the task with the higher degree of line damage will be prioritized.

[0046] Reference Figure 8 As shown, based on the inspection plan, the deployment of returning drones and drones in drone vehicles includes the following steps: Using the minimum number of drones calculation formula, calculate the number of drones required to execute the inspection plan, and use this as the number of features; Obtain the distance from the location of the inspection plan to the unmanned vehicle as the feature distance; Non-patrol drones that return to their destination at a distance less than the characteristic distance from the location of the inspection plan are designated as target drones, and the target drones are deployed to execute the inspection plan. The number of target drones is counted as the target count. The number of compensations is obtained by subtracting the target count from the number of features. Non-patrol drones in the unmanned vehicle are deployed to perform inspection plans, and the number of non-patrol drones in the deployed unmanned vehicle is equal to the number of compensation drones. The formula for calculating the minimum number of drones is as follows:

[0047] In the formula, I represents the number of features, S represents the severity of the abnormal warning corresponding to the inspection plan, and Z represents the number of non-patrol drones in the unmanned vehicle.

[0048] An adaptive patrol and inspection drone vehicle performs patrol and inspection operations according to the aforementioned adaptive patrol and inspection method.

[0049] This step is designed to match drones based on the severity of the inspection plan. When the situation is more severe, more drones are needed. To save power, drones should be deployed based on their distance. Since returning non-patrol drones have no missions, they can also accept missions. Therefore, when the distance between the returning non-patrol drone and the inspection plan is smaller than the distance between the drone vehicle and the inspection plan, the mission is assigned to the returning non-patrol drone, which saves more power.

[0050] Furthermore, this solution also proposes a storage medium on which a computer-readable program is stored, which executes the aforementioned adaptive patrol and inspection method when invoked.

[0051] It is understandable that the storage medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid-state drive (SSD).

[0052] In summary, the advantages of this invention are as follows: by predicting the inspection situation, different inspection plans are formed for inspections of different severity levels, and the execution order of the inspection plans is set according to the severity of the abnormality warnings corresponding to the inspection plans. This ensures that the number of drones matched with the abnormality is matched, and that more urgent abnormalities are prioritized for inspection. In addition, drones returning to base and drones in drone vehicles are deployed simultaneously, avoiding the situation where only drones in drone vehicles are deployed. This can reduce the overall flight distance of drones to a certain extent. At the same time, daily patrols are conducted in high-risk areas to further ensure the comprehensiveness of the inspection.

[0053] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. An adaptive patrol and inspection method, characterized in that, include: The drone vehicle is dustproof and off-road capable. It carries at least one drone, which communicates via satellite. The audio and video collected by the drone during inspections are transmitted back to the command center in real time. The number of drones in the unmanned vehicle is obtained, and based on historical data, the drones of the unmanned vehicle are divided into patrol drones and non-patrol drones; Based on the information collected by the UAV data acquisition system, the high-precision real-scene 3D map is modified in real time. Based on historical data of anomalies and high-precision real-scene 3D maps, a calculation model for high-risk areas is established. Obtain the actual data of the high-risk area required in the calculation mode of the high-risk area, substitute the actual data of the high-risk area into the calculation mode of the high-risk area, calculate the high-risk index of the high-risk area, set the patrol frequency of the high-risk area based on the high-risk index of the high-risk area, and patrol drones patrol the high-risk area. When the unmanned vehicle receives an anomaly warning, it uses AI technology to analyze the anomaly category of the warning and formulate an inspection plan accordingly. Establish a task severity assessment model, obtain the severity of anomaly warnings based on the task severity assessment model, and set the execution order of inspection plans based on the severity of the anomaly warnings corresponding to the inspection plans. The inspection plan is executed according to the order of execution. Based on the inspection plan, non-patrol drones returning to base and non-patrol drones of unmanned vehicles are deployed, and the deployed non-patrol drones execute the inspection plan.

2. The adaptive patrol and inspection method according to claim 1, characterized in that, The real-time modification of the high-precision real-scene 3D map based on information collected by the UAV data acquisition system includes the following steps: At least one image acquired by the UAV data acquisition system is used as a feature image; The acquisition location of the feature image is obtained. The image area in the high-precision real-scene 3D map that matches the acquisition location is used as the image to be compared. The feature image corresponding to the same acquisition location is paired with the image to be compared. The degree of change is calculated using the formula for the degree of change, and the rate of change is calculated using the formula for the rate of change. Replace the corresponding comparison image in the high-precision real-scene 3D map with the feature image whose degree of change is higher than the first preset value or whose rate of change is higher than the second preset value. The first and second preset values ​​are set based on empirical data. The formula for the degree of change is as follows: ； In the formula, C represents the degree of change, and I represents the number of pixels that differ between the image to be compared and the corresponding feature image. The formula for the rate of change is as follows: ； In the formula, V is the rate of change. The value is the median, and i is the index. Let represent the degree of change of the i-th feature image, and n be the total number of feature images.

3. The adaptive patrol and inspection method according to claim 2, characterized in that, The calculation model for establishing high-risk areas includes the following steps: Areas where anomalies occur more than the preset number of times are designated as high-risk zones. The number of anomalies occurring in high-risk areas is used as a risk reference factor; Calculate the potential hazard factor of a high-risk area using the potential hazard factor formula: Calculate the site hazard coefficient using the site hazard coefficient formula: Using the analytic hierarchy process (AHP), the weights of the hazard reference coefficient, the site hazard coefficient, and the potential hazard coefficient are obtained. The high-risk index of the high-risk area is then calculated using the high-risk index calculation formula. The formula for calculating the high-risk index is as follows: ； In the formula, Q is the high-risk index, t is the risk reference coefficient, o is the site risk coefficient, r is the potential risk coefficient, T is the weight of the risk reference coefficient, O is the weight of the site risk coefficient, and R is the weight of the potential risk coefficient. The formula for the potential hazard factor is as follows: ； In the formula, r is the potential hazard coefficient, p is the power line density in the high-risk area, f is the dead tree density in the high-risk area, and e is the natural constant. The formula for the site risk factor is as follows: ； In the formula, o is the site hazard coefficient, w is the vegetation cover density of the high-risk area, and s is the area in the high-risk area where the power line density is lower than the preset density.

4. The adaptive patrol and inspection method according to claim 3, characterized in that, Setting the patrol frequency for high-risk areas based on their high-risk index includes the following steps: The high-risk index of high-risk areas is rounded down to the nearest whole number and recorded as the whole index. The patrol frequency of high-risk areas is also recorded as the whole index. Based on the number of high-risk areas and the number of drones in the drone vehicle, the maximum patrol frequency is calculated using the formula for calculating the maximum patrol frequency. The rate of change of feature images containing high-risk areas is obtained as the feature change rate. When the rate of change of characteristics exceeds the preset rate, the patrol frequency of high-risk areas will be adjusted to the highest patrol frequency. The formula for calculating the maximum patrol frequency is as follows: ； In the formula, H represents the highest patrol frequency, N represents the number of high-risk areas, and L represents the number of patrol drones in the drone vehicle.

5. The adaptive patrol and inspection method according to claim 4, characterized in that, The process of using AI technology to analyze the anomaly categories of anomaly warnings and formulate inspection plans for them includes the following steps: Based on the number of abnormal power lines, the degree of abnormality, and the area covered by the abnormality, the anomalies in the historical anomaly warnings are classified to obtain at least one anomaly category. Extract at least one sample anomaly from the anomaly category. Based on historical data, obtain historical anomalies that are consistent with the sample anomalies. Use the union of the solutions to the historical anomalies as the inspection plan for the anomaly category. AI technology is used to analyze the anomaly categories of anomaly warnings, and the inspection plans corresponding to the anomaly categories of anomaly warnings are used as the inspection plans for anomaly warnings.

6. The adaptive patrol and inspection method according to claim 5, characterized in that, The establishment of the task severity assessment model includes the following steps: Based on the acquired task information, AI is used to analyze the number of abnormal power lines and the number of potential abnormal power lines, and the sum of the number of abnormal power lines and the number of potential abnormal power lines is recorded as the line hazard level. Based on the acquired task information, AI analysis is used to determine the extent of line damage and the geographical area where the line damage occurred. Using the analytic hierarchy process (AHP), we obtained the weights of the line hazard level, the line damage degree, and the geographical area of ​​the line damage. The severity of anomaly warnings is calculated based on the formula for calculating the severity of anomaly warnings. The formula for calculating severity is as follows: ； In the formula, s represents the severity, d represents the line hazard level, b represents the degree of line damage, m represents the geographical area where the line damage occurred, D represents the weight of the line hazard level, B represents the weight of the degree of line damage, and M represents the weight of the geographical area where the line damage occurred.

7. The adaptive patrol and inspection method according to claim 6, characterized in that, The severity of anomaly alerts, determined based on the task severity assessment model, includes the following steps: By substituting the actual data of the tasks for which anomaly warnings are required in the task severity assessment model into the model, the severity of the anomaly warnings can be obtained.

8. The adaptive patrol and inspection method according to claim 7, characterized in that, The process of setting the execution order of inspection plans based on the severity of the abnormal warnings corresponding to the inspection plans includes the following steps: Compare the severity of anomaly warnings and prioritize tasks with higher severity warnings. If the severity of the abnormal warnings for two tasks is the same, then compare the line hazard of the two tasks and prioritize the task with the higher line hazard. If the severity of the abnormal warning and the line hazard level of the two tasks are the same, then compare the area of ​​the line damage that occurred in the two tasks, and prioritize the task with the higher area of ​​line damage. If the severity of the abnormal warning, the degree of line danger, and the area of ​​line damage are the same for two tasks, then the task with the higher degree of line damage will be prioritized.

9. The adaptive patrol and inspection method according to claim 8, characterized in that, The deployment of drones for returning drones and drone vehicles based on the inspection plan includes the following steps: Using the minimum number of drones calculation formula, calculate the number of drones required to execute the inspection plan, and use this as the number of features; Obtain the distance from the location of the inspection plan to the unmanned vehicle as the feature distance; Non-patrol drones that return to their destination at a distance less than the characteristic distance from the location of the inspection plan are designated as target drones, and the target drones are deployed to execute the inspection plan. The number of target drones is counted as the target count. The number of compensations is obtained by subtracting the target count from the number of features. Non-patrol drones in the unmanned vehicle are deployed to perform inspection plans, and the number of non-patrol drones in the deployed unmanned vehicle is equal to the number of compensation drones. The formula for calculating the minimum number of drones is as follows: ； In the formula, I represents the number of features, S represents the severity of the abnormal warning corresponding to the inspection plan, and Z represents the number of non-patrol drones in the unmanned vehicle.

10. An adaptive patrol and inspection unmanned vehicle, performing patrol and inspection operations according to the adaptive patrol and inspection method as described in any one of claims 1-9.

Images