A power inspection multi-unmanned aerial vehicle cooperative task scheduling method and system

Abstract

The application discloses a kind of power inspection multi-unmanned aerial vehicle cooperative task scheduling method and system, it is related to unmanned aerial vehicle scheduling technical field.The application first obtains multiple inspection data, and establishes the multi-unmanned aerial vehicle cooperative task scheduling model including task uniqueness, task full coverage, unmanned aerial vehicle capability matching, maximum endurance time, energy consumption, task time window, track continuity, take-off and landing closure and return safety constraints;The importance of tower inspection task is determined by task priority quantization method, and the task announcement, unmanned aerial vehicle bidding, bid evaluation and task confirmation are completed based on the improved contract net protocol, and the unmanned aerial vehicle inspection sequence, inspection task point transfer flight path and return path are cooperatively optimized by combining the improved ant colony algorithm.The method of the application has lower total inspection cost, lower inspection task time, and can reduce the coincidence of unmanned aerial vehicle inspection path.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) scheduling technology, specifically to a method and system for scheduling collaborative tasks involving multiple UAVs for power line inspection. Background Technology

[0002] Power line inspection is an important technical means to ensure the safe and stable operation of power transmission lines. Traditional inspection methods mainly rely on manual on-site inspections along power transmission lines, tower areas, or corridors. This method is easily affected by terrain, weather, traffic conditions, the risks of working at heights, and the experience level of the inspection personnel, resulting in problems such as long inspection cycles, low work efficiency, delayed discovery of hidden dangers, and high personnel safety risks.

[0003] In recent years, drone inspection technology has been applied in the power industry. Drones can carry inspection payloads such as visible light cameras, infrared thermal imagers, lidar, and ultraviolet imagers to inspect transmission towers, insulators, fittings, conductors, ground wires, road obstructions, and the surrounding environment, thereby improving the mobility, automation, and data acquisition accuracy of inspection operations to a certain extent.

[0004] However, in large-scale power line inspection scenarios, a single drone is limited by its maximum endurance, flight radius, payload capacity, weather conditions, energy consumption, and return-to-home safety margin, making it difficult to cover all the poles to be inspected in one go. Especially when there are a large number of poles, a wide spatial distribution, and some poles have highly critical equipment or are subject to high urgency, relying on manual experience-based assignment or single-drone path planning can easily lead to uneven drone task allocation, overload of some drones, idle resources of some drones, overlapping flight paths, conflicting task execution sequences, insufficient return-to-home safety, and high overall operating costs.

[0005] Existing multi-UAV inspection task scheduling methods mainly include task allocation methods based on the shortest distance principle, path planning methods based on genetic algorithms, inspection path search methods based on ant colony algorithms, task clustering methods based on self-organizing mapping networks, and task negotiation methods based on contract network protocols. However, in large-scale power transmission line inspection scenarios, existing multi-UAV inspection task scheduling methods still have the following shortcomings: First, task allocation and flight path planning are usually processed independently in stages. The task allocation results are difficult to reflect the subsequent flight distance, turning time, path continuity, return path and airspace safety constraints in a timely manner, resulting in insufficient matching between the generated task sequence and the actual executable path. Second, some methods only use path distance or task completion time as optimization targets, without fully integrating factors such as the frequency of historical tower failures, equipment criticality, inspection urgency, task priority, task time window, UAV remaining energy, and flight energy consumption, making it difficult to meet the comprehensive scheduling needs of complex inspection tasks. Third, the existing task negotiation method is insufficient in comprehensively evaluating the status of drone resources, the importance of the task, the expected execution time, the expected energy consumption, and the current task load during the task release and drone bidding process. This can easily lead to problems such as invalid bids, duplicate negotiations, inefficient allocation, and untimely response to high-priority tasks. Fourth, when traditional ant colony algorithms, genetic algorithms, or single contract network protocols are used alone in large-scale tower inspection tasks, they have problems such as large search space, slow convergence speed, local optima, parameter sensitivity, and low efficiency of path replanning. Fifth, existing methods typically lack status feedback and dynamic rescheduling mechanisms during mission execution, making it difficult to dynamically adjust existing scheduling schemes and flight paths in a timely manner based on UAV location, remaining battery power, mission completion status, sudden failures, new inspection tasks, or changes in environmental conditions.

[0006] Therefore, it is necessary to propose a multi-UAV collaborative task scheduling method and system for power line inspection. This method comprehensively considers the spatial distribution of towers to be inspected, tower task priority, task time window, UAV performance status, endurance, flight energy consumption, flight path continuity, return safety, and dynamic changes in the operating environment. It forms a multi-UAV inspection scheduling scheme that optimizes task allocation and flight path coordination, thereby improving the task completion efficiency, resource utilization, flight path executability, and dynamic adaptability of multi-UAV power line inspection operations. Summary of the Invention

[0007] To address at least one of the aforementioned problems, this invention proposes a method and system for scheduling collaborative tasks involving multiple unmanned aerial vehicles (UAVs) for power line inspection.

[0008] The technical solution of this invention is a method for scheduling collaborative tasks involving multiple unmanned aerial vehicles (UAVs) for power line inspection, comprising the following steps: S1. Obtain multi-dimensional inspection data within the area to be inspected and preprocess it. The multi-dimensional inspection data includes power pole foundation data, inspection task requirement data, UAV performance status data, operating environment data, and historical inspection task data. S2. Based on the preprocessed multivariate inspection data, a multi-UAV collaborative inspection task scheduling model is established for collaborative optimization of the task allocation results and flight trajectories of multiple UAVs. The multi-UAV collaborative inspection task scheduling model includes decision variables, objective function and constraints. S3. Based on the improved contract network protocol, before the mission announcement is released, candidate drones are screened according to the aforementioned constraints. During the bidding stage, only candidate drones that meet the requirements of mission capability, expected execution time, expected energy consumption, mission time window, and return safety are allowed to generate bidding responses. The winning drone is determined by combining the bidding utility value. S4. Based on the improved ant colony algorithm, before each state transition in constructing the candidate scheduling scheme, a set of candidate inspection task points and a set of candidate flight segments are formed according to the constraints. Candidate inspection task points or candidate flight segments that do not meet the constraints are eliminated. The inspection sequence of the UAV, the transfer trajectory between inspection task points and the return path are optimized, and the scheduling scheme is output.

[0009] Another objective of this invention is to disclose a multi-UAV collaborative task scheduling system for power line inspection, comprising: The input and preprocessing module is configured to acquire multi-source inspection data and preprocess it. The inspection task modeling module is configured to use preprocessed multivariate inspection data to establish a multi-UAV collaborative inspection task scheduling model, objective function, and constraints. The scheduling task output module is configured to generate a scheduling scheme using an improved contract net protocol and an improved ant colony algorithm; The drone swarm management module is configured to send the scheduling scheme to the corresponding drones and control multiple drones to perform take-off, inspection and return according to the scheduling scheme.

[0010] The beneficial effects of this invention are as follows: First, this invention utilizes an improved contract network protocol to determine the winning drone, and then plans the inspection routes of the winning drone based on an improved ant colony algorithm, thereby generating a scheduling scheme. Compared to traditional scheduling scheme generation methods, this invention's method has lower total inspection costs, lower inspection task time, and reduces drone inspection path overlap. It can improve the scheduling efficiency, path rationality, task responsiveness, and dynamic adaptability of multi-drone collaborative operations in complex power inspection scenarios. Attached Figure Description

[0011] Figure 1 A schematic diagram of the overall process for scheduling multi-drone collaborative tasks in power line inspection; Figure 2 A schematic diagram of a multi-UAV mission scheduling and flight trajectory collaborative optimization framework; Figure 3 A two-dimensional horizontal projection and candidate flight segment diagram for a multi-UAV collaborative inspection mission; Figure 4 A schematic diagram illustrating the turning angle and flight path constraints of a UAV in continuous flight segments; Figure 5 A flowchart for multi-UAV mission negotiation and allocation based on the improved ContractNet protocol; Figure 6 A flowchart for collaborative optimization of mission scheduling and flight path based on the improved CNP-ACO; Figure 7This is a schematic diagram of the task execution status feedback and dynamic rescheduling process; Figure 8 A schematic diagram showing the results of task allocation and flight path planning for multi-UAV collaborative inspection; Figure 9 A comparison chart of task completion times under different scheduling methods; Figure 10 A comparison chart of average task allocation success rates under different scheduling methods; Figure 11 A comparison chart showing the average generation time of scheduling schemes under different scheduling methods. Detailed Implementation

[0012] The specific embodiments of the present invention will be clearly and completely described below with reference to examples. Obviously, the described examples are only some embodiments of the present invention, and not all embodiments.

[0013] Example 1: As Figure 1 , Figure 2 As shown, a method for scheduling collaborative tasks involving multiple UAVs for power line inspection includes the following steps: S1. Obtain multi-dimensional inspection data within the area to be inspected and preprocess it. The multi-dimensional inspection data includes power pole foundation data, inspection task requirement data, UAV performance status data, operating environment data, and historical inspection task data. Specifically, in this embodiment, the basic data of the power poles includes the pole number to be inspected, two-dimensional horizontal coordinates, spatial coordinates, task height, line affiliation, equipment type, connection relationship between adjacent poles, and spatial distribution relationship.

[0014] Among them, the tower number to be inspected serves as a unique identifier for the tower to be inspected and its corresponding inspection task. It is used to associate and store the tower's coordinates, task height, equipment type, priority, task allocation results, and track access order. It can also be temporarily coded.

[0015] The spatial position of the inspection task point corresponding to the tower to be inspected is represented by two-dimensional horizontal coordinates and task height. The two-dimensional horizontal coordinates represent the position of the tower to be inspected within the horizontal projection plane, while the task height represents the vertical position of the corresponding inspection task point. To reduce the computational complexity of multi-UAV task allocation and inspection sequence optimization, during model building, the horizontal position of each tower to be inspected is projected onto a two-dimensional plane, and the task height is retained as an independent vertical attribute associated with the corresponding two-dimensional horizontal coordinates. Therefore, the two-dimensional horizontal coordinates represent the horizontal projection of the three-dimensional spatial position of the inspection task point, but are not a complete replacement for the three-dimensional spatial coordinates.

[0016] The two-dimensional horizontal coordinates are used to calculate the horizontal distance from the current position of the UAV to the inspection task point and between different inspection task points, and to determine the spatial proximity relationship of candidate inspection task points; the horizontal distance is further used together with the vertical height difference between inspection task points to calculate the spatial transfer distance, estimated flight time, estimated energy consumption and track cost between inspection task points.

[0017] Let the first The inspection task points corresponding to the towers to be inspected are: In the formula, Indicates the first The inspection task points corresponding to the towers to be inspected; and These represent the x-coordinate and y-coordinate of the inspection task point in the two-dimensional horizontal projection plane, respectively. This indicates the task height of the inspection point; This refers to the inspection task point number corresponding to the tower to be inspected. This represents the total number of poles or inspection points to be inspected. The two-dimensional horizontal coordinates and the task height serve as the horizontal and vertical position attributes of the inspection task points, respectively. These are stored in association with the corresponding pole numbers to be inspected and together represent the three-dimensional spatial position of the corresponding inspection task points.

[0018] Line attribution refers to the transmission line or line section to which the tower to be inspected belongs. It is used to establish the topological relationship between each tower to be inspected on the same transmission line and to help determine the task connection relationship and candidate routes during the continuous line inspection process.

[0019] Equipment type refers to the types of sub-tasks on the tower to be inspected, including the tower body, number of insulators, types and quantities of hardware, conductors, grounding devices, etc. In actual use, these equipment types can be numbered, and the inspection time, inspection load requirements, and task energy consumption of each type of equipment can be recorded.

[0020] The connection relationships between adjacent towers are used to establish the topological adjacency matrix between the towers to be inspected: In the formula, This represents the topological adjacency matrix between the towers to be inspected; Representation matrix The Middle line, number The elements of the column are used to represent the first... The inspection task point and the first Do the inspection task points have a direct topological connection relationship? ,and When the inspection task point Inspection task points When the corresponding towers belong to the same transmission line and are physically adjacent to each other... ;otherwise, The topological adjacency matrix is ​​used to subsequently determine the preferred candidate flight segments in the continuous route inspection process, judge the topological continuity between inspection task points, and assist in forming a set of candidate inspection task points.

[0021] The spatial distribution relationship is used to establish a set of neighboring tasks for each tower to be inspected based on its two-dimensional horizontal coordinates, the distance between towers, and the proximity of the area. In the formula, Indicates the inspection task point The corresponding set of nearby inspection task points; Indicates the first Each inspection task point; This represents the set of inspection task points corresponding to all poles to be inspected; condition Indicates the inspection task point Inspection task points Directly adjacent in the transmission line topology. The aforementioned set of neighboring tasks is used for subsequent candidate inspection task point selection, spatial proximity determination of inspection task points, and inspection sequence optimization.

[0022] The data required for inspection tasks include inspection type, historical failure frequency, equipment criticality, inspection urgency, task time window, and task completion requirements.

[0023] The inspection type indicates that different inspection methods are required for different equipment types, such as visible light appearance inspection, infrared temperature measurement inspection, ultraviolet discharge detection, lidar scanning, local re-inspection, and channel environment inspection. It is difficult to integrate so many inspection methods into a single inspection drone. Moreover, the time and energy consumption of different inspection methods are also different. Therefore, in actual work, it is necessary to number these inspection methods and mark their specific inspection time and energy consumption.

[0024] Historical fault frequency is used to characterize the frequency of faults or anomalies occurring on the towers or corresponding equipment to be inspected within a preset historical statistical period. The historical fault frequency is determined based on fault records, anomaly alarm records, or historical inspection results stored in the power equipment operation and maintenance management system.

[0025] Specifically, the poles to be inspected are designated. Or the number of times the equipment corresponding to the inspection task has experienced a failure or abnormality within a preset historical statistical period. The length of the preset historical statistical period is Then the towers to be inspected Historical failure frequency for: In the formula, Indicates the first Historical fault frequency of each tower to be inspected or its corresponding equipment; Indicates the first The total number of times a pole or its corresponding equipment to be inspected has experienced a failure or abnormality within a preset historical statistical period, of which the superscript indicates... Indicates a fault or abnormal event; This indicates the length of the preset historical statistical period, where the subscript... Indicates the historical statistical period; This indicates the number of the pole or inspection point to be inspected. Historical fault frequency. The unit is composed of The time unit determines this; for example, when When measured in years, The unit is times / year.

[0026] When all towers to be inspected use the same historical statistical period, the number of faults or anomalies occurring within the preset historical statistical period can be directly used as the characterization value of historical fault frequency. The higher the historical fault frequency, the more frequently faults or anomalies occurred during the historical operation of the tower or corresponding equipment to be inspected, and the higher the priority of the corresponding inspection task.

[0027] Equipment criticality is used to characterize the degree of impact of the corresponding equipment on the safe operation of the line and the reliability of power supply, and criticality scores can be given for each piece of equipment.

[0028] The urgency level of the inspection is used to characterize the degree to which the current task needs to be performed in a timely manner; The task time window is used to limit the time range during which a task is allowed to start and complete.

[0029] Among them, historical failure frequency, equipment criticality and inspection urgency are used to calculate tower task priority; task time window is used to limit the time range within which inspection tasks can start and be completed, and to help determine the order of task announcements; inspection type and task completion requirements are used to determine inspection load requirements and task execution conditions.

[0030] The drone performance status data includes the number of drones, current location, maximum flight time, flight speed, turn rate, minimum turning radius, payload capacity, remaining battery power, and availability.

[0031] Load capacity is used to characterize its ability to carry different inspection methods.

[0032] The availability status indicates whether a drone can participate in task allocation during the task period corresponding to the inspection task to be performed. Specifically, a drone is marked as available when it is in normal online operation, has normal communication, is not malfunctioning, is not under maintenance, and is not occupied by other tasks that conflict with the current inspection task; otherwise, it is marked as unavailable. When screening candidate drones based on the improved contract network protocol, only drones in the available status enter the subsequent candidate qualification screening, and feasibility is further judged based on conditions such as remaining battery power, maximum flight time, payload capacity, and expected energy consumption.

[0033] The operational environment data includes wind speed, wind direction, terrain environment, obstacle information and no-fly zone information. Wind speed and wind direction can be used to calculate the effective flight speed of the drone. When flying with the wind, the wind speed is positive and when flying against the wind, the wind speed is negative. Flight speed, turn rate, minimum turning radius, and payload capacity are all inherent capabilities of the drone itself; while remaining battery power takes into account whether the drone can continue to perform the next task after completing the current one.

[0034] After obtaining the above data, preprocessing is required. In this step, to facilitate subsequent calculations, the specific locations of the towers to be inspected must first be determined. Therefore, in this embodiment, to reduce the computational complexity of multi-UAV task allocation and inspection sequence calculation, the horizontal positions of each tower within the inspection area are projected onto a two-dimensional plane to establish a two-dimensional horizontal projection task scene. The two-dimensional horizontal projection task scene is used to represent the horizontal position, adjacent connection relationship, and spatial distribution relationship of each inspection task point; the task height of each inspection task point is retained as an independent vertical attribute associated with the corresponding horizontal position and is not discarded due to horizontal projection processing.

[0035] To facilitate the planning of UAV inspection routes, obstacles and no-fly zones in the operational environment data are marked within the planar projection range of the two-dimensional horizontal projection task scenario. These areas are then avoided during subsequent horizontal route planning. The final result is as follows: Figure 3 As shown.

[0036] Let the set of towers to be inspected be . , No. One pole tower awaiting inspection Corresponding inspection tasks and inspection task points Initial set of all inspection task points. The task points that have not yet been inspected constitute the set of task points to be inspected. .

[0037] Meanwhile, to facilitate the establishment of a subsequent task allocation model, the drones participating in this collaborative inspection task are represented in a set, establishing a drone ID set and a drone set. , in, This indicates the assembly of towers to be inspected. Indicates the first One pole tower awaiting inspection; This indicates the corresponding inspection task; Indicates the first The coordinates of the inspection task points corresponding to the towers to be inspected; This represents the collection of all inspection task points; This represents the set of inspection points that have not yet been inspected; Represents a set of drone IDs; This refers to the collection of drones participating in this collaborative inspection mission; Indicates the first A drone; This indicates the total number of poles or inspection points to be inspected. This indicates the total number of drones involved in task allocation; , .

[0038] S2. Based on the preprocessed multivariate inspection data, establish a multi-UAV collaborative inspection task scheduling model for collaborative optimization of task allocation results and flight trajectories of multiple UAVs. The multi-UAV collaborative inspection task scheduling model includes decision variables, objective function and constraints.

[0039] Based on the preprocessed multivariate inspection data, the model parameters and decision variables of the multi-UAV collaborative inspection task scheduling model are determined.

[0040] Based on the two-dimensional horizontal coordinates, mission altitude, connection relationships between adjacent towers, and spatial distribution relationships of each tower to be inspected, the horizontal distance, vertical height difference, and spatial proximity relationships between the towers to be inspected are calculated, and a set of candidate flight segments is established by combining the topological adjacency matrix. Furthermore, by combining the UAV flight speed, wind speed and direction in the operational environment, obstacle information, and no-fly zone information, the estimated transfer time, estimated flight energy consumption, and track reachability for each candidate flight segment are calculated. The horizontal distance, vertical height difference, estimated transfer time, estimated flight energy consumption, and track reachability constitute the flight segment parameters of the scheduling model.

[0041] Based on the set of poles to be inspected, the set of drones, and the set of feasible flight segments, decision variables are set, including task allocation variables and flight segment selection variables. The task allocation variable characterizes the task allocation relationship between the poles to be inspected and the drones, while the flight segment selection variable characterizes whether the drone selects a candidate flight segment between the corresponding inspection task points when performing an inspection task. If necessary, drone activation variables, task start and completion time variables, cumulative execution time variables, and cumulative energy consumption variables are also set and associated with the task allocation variables and flight segment selection variables.

[0042] The set of feasible flight segments is used for flight segment selection, comprehensive trajectory cost calculation, trajectory continuity constraints, and candidate inspection task point screening in the task scheduling model.

[0043] In this step, the objective function of the multi-UAV collaborative inspection and scheduling model is as follows: In the formula, Indicates the first One candidate multi-UAV collaborative inspection scheduling scheme, including inspection task allocation results, UAV activation status and flight segment selection results; Indicate candidate scheduling schemes The smaller the comprehensive objective function value, the better the overall performance of the scheduling scheme. Assign completeness metrics to tasks; This is an indicator of drone resource utilization. Considering the overall cost of the flight path; Delay penalties for high-priority tasks; The coefficient of variation for each UAV mission payload; The smaller the value, the more balanced the number of tasks or workload of each drone. , , , and These represent the weighting coefficients for task allocation integrity, UAV resource utilization, overall trajectory cost, high-priority task delay penalty, and task load balancing metrics, respectively. .

[0044] The formulas for calculating each parameter in the objective function are as follows: ; ; , , ; ; ; , , ; , , , , , , , , .

[0045] In the formula, This indicates the total number of inspection task points corresponding to the towers to be inspected; This indicates the total number of drones involved in task allocation; Indicates the inspection task point number; Indicates the drone's serial number; This indicates an indicator function that takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. Indicates the first Should drones be used to inspect poles awaiting inspection? If so, then It is 1 if it is true, otherwise it is 0. Indicates the first The active state variable of the drone, if the first If each drone is assigned at least one inspection task, then ,otherwise ; Representing the scheduling scheme Total flight distance of all drones in the region; This represents the maximum reference total flight distance of all available UAVs, used to normalize the total flight distance of the scheduling scheme; Representing the scheduling scheme Total projected energy consumption for all drones in the region; This represents the maximum reference available energy used for energy consumption normalization; Indicates the first The drones were deployed according to the scheduling plan The estimated total energy consumption when performing all assigned tasks; Represents a node To the node The feasible flight segment distance between them that meets obstacle avoidance and no-fly zone restrictions; Indicates the first Whether the drone executes directly depends on the node. fly to node The segment of the voyage; when The time indicates that the flight segment has been selected; otherwise... ; Indicates the first Reference cruising speed when a drone is performing an inspection mission; Indicates the first The nominal maximum endurance of the drone under standard operating conditions is used to calculate the distance normalization parameter. ; Indicates the first Is the drone currently available? and These represent the flight distance cost weight and energy consumption weight, respectively. + =1, in specific settings, if more emphasis is placed on flight distance cost, then Larger, and vice versa; Indicates the first The overall priority weight of each inspection task point; This indicates the threshold for determining high-priority tasks. At that time, the first The inspection task was identified as a high-priority task. Indicates the first The actual start time of each inspection task; Indicates the first The latest possible or expected start time for each inspection task; This represents the maximum reference time used to normalize task delay times; This means that the penalty value is 0 when there is no delay in the task, and a delay penalty is only incurred when the actual start time is later than the latest start time. Indicates the first Each inspection task corresponds to the historical failure frequency of the equipment. This indicates the critical score for the equipment; This indicates the urgency level of the inspection task. and These represent the minimum and maximum values ​​of historical fault frequency among all tasks to be inspected; and These represent the minimum and maximum values ​​of the equipment criticality score, respectively. and These represent the minimum and maximum values ​​of the task urgency score, respectively; , and These represent the historical failure frequency, equipment criticality, and task urgency after minimum-maximum normalization. Indicates the first The overall priority score for each inspection task; , and These represent the weighting coefficients corresponding to historical failure frequency, equipment criticality, and task urgency, respectively. Indicates the first The number of tasks undertaken by each drone; This indicates the average number of tasks assigned to each drone participating in task allocation.

[0046] The constraints are as follows: Binary task assignment variable ,in: Task uniqueness constraint: This constraint condition means that for any pole to be inspected... Only one drone can be assigned to a pole to prevent the same pole from being assigned to the same drone repeatedly. Full task coverage constraint: This constraint condition means that for any pole to be inspected... At least one drone should be assigned to a task to prevent it from being missed. When the above two constraints work together, we have: This indicates that each pole to be inspected is inspected by exactly one drone.

[0047] Unmanned aerial vehicle (UAV) capability matching constraints: This constraint means that only when That is, the first The drone meets the first Only when the task requires certain capabilities is it permitted. When the first The inspection payload type, payload capacity, and availability status of the drone all meet the requirements of the first... When the requirements of an inspection task are met, order Otherwise, .

[0048] Energy consumption constraints: , In the formula, This represents the maximum reference available energy used for energy consumption normalization. Indicates the first The upper limit of the actual usable energy of a drone; Indicates the first The theoretical maximum available battery energy for a drone; Indicates the first The available energy deviation coefficient of the drone's battery; This indicates the maximum permissible energy deviation percentage.

[0049] Task time window constraints: In the formula, Indicates the inspection task point The earliest time when inspections are permitted to begin; Indicates the inspection task point The actual start time of the inspection; Indicates the inspection task point The actual time when the inspection was completed; Indicates the inspection task point The latest time allowed to complete the inspection.

[0050] Track continuity constraints: In the formula, Indicates the first A drone was deployed to the inspection task point. The number of flight segments; Indicates the first The drone departed from the inspection task point. The number of flight segments; when At that time, the first The drone must arrive at the inspection task point by precisely passing through an entry flight segment. And leave the inspection task point via a departure segment. .like Figure 4 As shown.

[0051] Takeoff and landing closure constraints: In the formula, Indicates the first The drone's nest, and: ; Indicates the first Whether the drone is from its own nest node Fly directly to the inspection mission point ; Indicates the first Is the drone dispatched from the inspection task point? Return directly to the host node .when At that time, the first Each drone must depart from the nest once and return to the nest once after completing its mission; when At this time, the drone does not generate takeoff, inspection, and return flight segments.

[0052] Return-to-base safety constraints: This constraint indicates and They represent the first Drones complete inspection tasks Cumulative energy consumption and cumulative execution time; and These respectively indicate the drone's location from the inspection task point. The estimated energy and estimated time required to return to the nest along a feasible return path that meets obstacle avoidance and no-fly zone restrictions; and These represent the preset safe return energy margin and safe time margin, respectively; The default value is a large positive number; Indicates the first The upper limit of the actual usable flight time of a drone, taking into account battery status, weather conditions, and safety margins. This constraint ensures that the drone has the energy and time to safely return to its corresponding nest after completing any assigned task. This invention employs a comprehensive optimization objective to evaluate the scheduling scheme. Specifically, task allocation completeness is used to evaluate whether all poles to be inspected have been uniquely allocated; UAV resource utilization rate is used to evaluate UAV resource usage; comprehensive trajectory cost is used to evaluate flight distance and energy consumption; high-priority task delay penalty is used to reduce response time for important inspection tasks; and task load balancing is used to improve the efficiency of multi-UAV collaborative operations. Each optimization indicator is assigned a corresponding weight based on the specific inspection scenario to achieve comprehensive optimization.

[0053] S3. Based on the improved contract network protocol, before the mission announcement is released, candidate drones are screened according to the aforementioned constraints. During the bidding stage, only candidate drones that meet the requirements of mission capability, expected execution time, expected energy consumption, mission time window, and return safety are allowed to generate bidding responses. The winning drone is determined by combining the bidding utility value. In this step, improving the contract network protocol involves refining the protocol itself. The contract network protocol typically includes steps such as task announcement, bidding, tender evaluation, contract award, task execution, and result feedback. Figure 5 As shown, this is a conventional method in the field, and its specific steps will not be described in detail here.

[0054] The improvement in this embodiment is that power inspection constraints and trajectory feedback are inserted into the task announcement, bidding, bid evaluation, and bid confirmation processes of the contract network protocol. Specific improvements are shown in Table 1. Table 1. Improvements to the Contract Network Protocol This embodiment improves the conventional contract network protocol's "task announcement - bidding - bid evaluation - bid award" process into a "capability pre-screening - priority announcement - constraint bidding - comprehensive utility bid evaluation - constraint conflict feedback and reallocation" process. In this process, the constraints directly affect the generation of candidate UAVs, candidate inspection task points, and candidate flight segments, without a separate post-verification step.

[0055] In this step, an improved contract network protocol is introduced during the task allocation phase, with the dispatch center acting as the bidding agent and each drone as the tendering agent. During the task allocation phase, the dispatch center acts as the bidding agent, each drone as the tendering agent, and each tower to be inspected is treated as an inspection task unit. Based on the priority weight and time window of each inspection task, the dispatch center cyclically issues task announcements for unassigned tasks in priority order. Each task announcement includes at least the coordinates of the tower to be inspected, the inspection type, the task priority, the task time window, and the load requirements.

[0056] Candidate drones generate bidding responses based on their current location, remaining battery power, payload capacity, currently assigned task set, current task load, estimated flight time, and estimated energy consumption. The dispatch center determines the winning drone for the current inspection task based on the bidding utility value and updates the assigned task set and the task set for each drone. It then continues to issue the next unassigned task until all pending inspection tasks have been initially assigned. The same drone can obtain multiple inspection tasks in different rounds of bidding. The winning drone receives the initial assignment result for the corresponding inspection task but does not immediately execute the task until collaborative trajectory optimization is completed. After all pending inspection tasks form an initial task allocation scheme, and the improved ant colony algorithm completes the collaborative optimization of the inspection order of each drone, the transfer trajectory between inspection task points, and the return path, the dispatch center issues the final scheduling scheme to the corresponding drone for execution.

[0057] The utility function used in this step is as follows: In the formula, Indicates the drone's serial number; Indicates the first Drones perform inspection tasks The bidding utility value; Indicates the inspection task point Priority level or priority score; This represents the maximum possible value for task priority, used for normalization. Indicates the first The drone carried out the first The estimated execution cost of each mission point can be determined by the estimated flight distance, mission execution time, mission energy consumption, and payload consumption. This represents the baseline track cost or baseline execution cost, used for normalization. Indicates task priority and reward weight; This represents the task execution cost weight, and can be set. .

[0058] In practice, the scheduling center selects the drone with the higher utility value as the winning drone. After the initial allocation of all pending inspection tasks is completed, based on the initial task set corresponding to each drone, an improved ant colony algorithm is invoked to optimize the collaborative flight paths of multiple drones. The improved ant colony algorithm constructs complete flight paths for each drone within the same candidate scheduling scheme, starting from its corresponding take-off and landing node, sequentially visiting its inspection task points, and returning to the take-off and landing node. It jointly optimizes the task access order, transfer segments between inspection task points, and return paths for each drone, and evaluates the candidate scheduling schemes based on task coverage, flight distance, task execution time, energy consumption, high-priority task delay, and load balancing.

[0059] The improved ant colony algorithm refers to improvements made to the following parts of the ant colony algorithm: In the ant construction phase, one ant is assigned to one multi-UAV cooperative scheduling scheme; in the state transition probability calculation phase, a composite heuristic factor including tower task priority, node distance, estimated time cost, estimated energy consumption, and constraint feasibility is added; in the next inspection task point selection phase, a candidate node filtering module is inserted; in the pheromone update phase, pheromone is updated according to the objective function; for infeasible schemes formed during the candidate scheduling scheme construction process, infeasible solutions are fed back to the improved contract network protocol. See Table 2 for details.

[0060] Table 2. Improvements to the Ant Colony Algorithm Specifically, the improved ant colony algorithm is as follows: Each ant corresponds to a complete multi-UAV candidate collaborative scheduling scheme. The candidate collaborative scheduling scheme includes the allocation relationship between the inspection task and the UAV, the task access order of each UAV, the transfer segment between inspection task points, and the return path from the last inspection task point back to the corresponding take-off and landing node.

[0061] Composite heuristic factor: in: Feasibility constraints: in, Indicates the first The node number where the drone is currently located. This indicates the number of the candidate next inspection point. Indicates the drone's serial number; and These represent the current inspection task point and the candidate next inspection task point, respectively. Indicates the first A drone from the node Transfer to node Composite heuristic factors; Indicates the distance between nodes; Indicates the first The overall priority weight of each candidate inspection task point; This indicates the maximum task weight; Indicates the estimated time cost; Indicates the expected energy consumption; Indicates the constraint feasibility factor; These represent the weighting coefficients corresponding to distance, task priority, time cost, and energy consumption, respectively; the task priority parameter in the composite heuristic factor uses the task points to be inspected. Task priority weight The task priority weight According to the poles to be inspected The task priority weight is calculated based on historical failure frequency, equipment criticality, and inspection urgency. Higher historical failure frequency, higher equipment criticality, or higher inspection urgency correspond to higher task priority weights. The larger.

[0062] Therefore, the historical failure frequency is weighted by task priority. By incorporating a composite heuristic factor and further exploring the state transition probability of the next inspection point for the UAV, the towers to be inspected with a high historical failure frequency are given a higher priority access probability.

[0063] Let the first The drone is currently located at node Then it selects the next inspection task point. The probability is: , in, Indicates the first In the algorithm iteration, the first Only one ant completed the first The set of inspection task points that have not yet been assigned after the next state transition; Indicates the first In the nth iteration, the 1st The drone is currently located at node At that time, the inspection task points will be... The state transition probability of selecting the next access node; Indicates the number of algorithm iterations; Indicates the current inspection task point number; Indicates the candidate next inspection task point number to be selected; This represents the traversal index in the candidate node set; Indicates the first Next iteration time segment The concentration of pheromones on the surface; Indicates the first A drone from the node Transfer to node Composite heuristic factors; Indicates the first A drone is located at the node At that time, under the constraints of load, endurance, remaining energy, mission time window, track continuity, and return safety, the set of candidate inspection mission points that can be accessed, and That is, a candidate point must first be a task point that has not yet been assigned in the current candidate scheduling scheme; Indicates the ant number; Indicates the first The state transition sequence of an ant when constructing a candidate scheduling scheme in the current iteration. This indicates the degree to which pheromone concentration affects the probability of node selection; This indicates the degree of influence of the composite heuristic factor on the node selection probability. The denominator is used to normalize the selection tendency of all feasible candidate nodes to ensure that the sum of the corresponding selection probabilities is 1.

[0064] In this embodiment, the constraint feasibility determination is not an independent constraint feasibility determination performed separately after the scheduling scheme is generated, but rather the constraint conditions established by S3 are directly applied during the candidate scheduling scheme construction process. Specifically, when the k-th UAV is at the current inspection task point... Select the next inspection task point Previously, the calculations included the inspection task points. The payload matching status after being added to the current task sequence, cumulative task execution time, estimated total energy consumption, task start and end times, inter-node track reachability, and return energy reserve after task completion are all considered. When the above parameters satisfy the UAV capability matching constraints, maximum endurance constraints, energy consumption constraints, task time window constraints, track continuity constraints, and return safety constraints, the inspection task points will be checked. Add the k-th UAV to the candidate inspection task point set; if any constraint is not satisfied, remove the inspection task point. They are removed from the candidate inspection task point set and are not included in the current state transition probability calculation.

[0065] During the construction of candidate scheduling schemes, the allocated task set ensures that the same tower to be inspected is not repeatedly assigned, and the completion of the assignment of all towers to be inspected is taken as the termination condition for scheme construction, thereby satisfying the constraints of task uniqueness and full task coverage. If each local inspection task point satisfies the constraints, but the overall trajectory formed by the combination does not meet the constraints due to task connection, cumulative energy consumption, time window conflict, or insufficient return safety margin, one or more infeasible tasks will be fed back to the improved contract network protocol, and candidate UAVs will be re-selected and locally reallocated.

[0066] Therefore, in this embodiment, the feasible solution space is limited by constraints during the scheduling scheme generation process, rather than using another set of conditions to make independent judgments after the scheduling scheme is generated.

[0067] After one iteration, update the pheromone levels based on pheromone evaporation and optimal track contribution: in, Indicates the number of algorithm iterations; Indicates the first After the next iteration, the flight segment The concentration of pheromones on the surface; Indicates the first The pheromone concentration of this flight segment after the next iteration; Indicates the pheromone evaporation coefficient; This indicates the optimal path pair segment in this iteration. The increase in pheromones; This represents the set of all directed segments included in the optimal cooperative scheduling scheme; This represents the comprehensive cost of the optimal collaborative scheduling scheme; Indicates the first The optimal cooperative scheduling scheme obtained in the next iteration includes the set of flight segments. If the flight segments If the optimal scheduling scheme is adopted, the pheromone level is increased by the reciprocal of its comprehensive cost value; otherwise, the pheromone increment for that flight segment is 0.

[0068] The improved CNP-ACO optimization process in this embodiment is as follows: Figure 6As shown, this method not only aims for the shortest flight distance but also considers task execution time and energy consumption as important factors in trajectory evaluation. When a candidate trajectory has too many turns, insufficient return energy, exceeds the time window, or consumes more energy than the UAV's available battery power, the system reduces the probability of selecting that trajectory or resubmits the relevant task to the improved contract network for local reallocation. Compared with conventional ant colony algorithms that construct heuristic information based solely on node distance and optimize only a single UAV's single access path, this embodiment encodes an ant into a complete collaborative scheduling scheme that includes the task allocation results of multiple UAVs, the task access order, the transfer trajectory between inspection task points, and the return path; it introduces a composite heuristic factor that includes task priority, segment distance, task execution time, energy consumption, and constraint feasibility; it filters candidate nodes that do not meet the requirements for payload, endurance, energy, time window, and return safety before state transition; and when a candidate trajectory is infeasible, it feeds back the relevant task to the improved contract network task allocation module for local reallocation, thereby forming a collaborative feedback mechanism between task allocation and trajectory optimization.

[0069] In actual dispatch and inspection processes, unexpected events often occur, such as: additional inspection tasks, wind speed, wind direction, drone malfunctions, or at least one drone flight path being blocked. Therefore, when facing these unexpected events, such as... Figure 7 As shown, the embodiment of the present invention also includes a rescheduling step: based on the sudden event and initial conditions, the constraints are regenerated, and a new scheduling scheme is generated by reusing the improved contract net protocol and the improved ant colony algorithm.

[0070] To further illustrate the superiority of the embodiments of the present invention, specific test examples are given below.

[0071] A simulation scenario was established using a 220 kV transmission line inspection area in Southwest China as the background. The inspection area was set as a rectangular airspace of 10 km × 10 km, with the terrain mainly consisting of plains. Four fixed fully automated UAV nests were set up within the area for the autonomous launch, retrieval, and task scheduling of the UAVs. There were 50 transmission towers to be inspected within the area, with each tower corresponding to a unique inspection task point. The tasks were divided into four categories based on historical fault frequency, equipment criticality, inspection urgency, and task time window: ordinary, critical, historically frequently faulty, and urgent.

[0072] The experiment used nine identical small multi-rotor inspection drones. Five of these drones were equipped with high-definition visible light cameras for routine appearance inspections, while the other four were equipped with dual-light (visible light and infrared) cameras for detailed inspections and troubleshooting. The drones' horizontal cruising speed was set to 12 m / s, maximum turning speed to 30° / s, maximum endurance to 35 min, and fixed cruising altitude to 25 m. The planned hovering inspection time for ordinary poles was set to 1 min, and for critical poles, it was set to 2 min.

[0073] The simulation was performed in MATLAB R2023b environment, with a computer configuration of Intel Core i7-13700H processor and 32 GB of memory. To reduce the impact of random ant colony initialization and task bidding process on the experimental results, each group of experiments was run independently 10 times under the same scene parameters and UAV configuration, and the average of the 10 run results was taken as the final evaluation result.

[0074] In the task scheduling performance verification, the method of this embodiment, the consensus-based binding algorithm CBBA, the cooperative backtracking genetic algorithm CBGA, and the traditional split CNP+ACO algorithm are compared; in the trajectory planning performance verification, the method of this embodiment, the CBBA-RRT algorithm, the CBGA-A* algorithm, and the traditional ACO algorithm are compared.

[0075] The final result is as follows Figures 8-11 As shown.

[0076] Figure 8 This is a schematic diagram of the multi-UAV collaborative inspection task allocation and trajectory planning results obtained by the method in this embodiment.

[0077] Figure 9 In the figure, the horizontal axis represents different experimental scenarios, and the vertical axis represents task completion time. As can be seen from the figure, the method in this embodiment, by coordinating task allocation, trajectory optimization, and dynamic rescheduling, can reduce UAV waiting, detours, and repeated scheduling, thereby improving the overall execution efficiency of multi-UAV inspection tasks.

[0078] Figure 10 and Figure 11 In the task allocation success rate ( Figure 10 ), trajectory planning time ( Figure 10 The performance of different scheduling methods is comprehensively evaluated from multiple dimensions, including task coverage, resource utilization, track continuity, task completion efficiency, and dynamic adaptability. The comparison results show that the method in this embodiment of the invention has a good balance in terms of task coverage, resource utilization, track continuity, task completion efficiency, and dynamic adaptability, and can improve the overall feasibility and robustness of multi-UAV collaborative inspection task scheduling schemes.

[0079] The task allocation success rate refers to the proportion of all pole inspection tasks to which a single scheduling calculation successfully allocates tasks to drones that meet constraints such as payload capacity, endurance, remaining battery power, task time window, flight path continuity, and return-to-home safety, without any omissions or duplicate allocations. The calculation formula is as follows: In the formula, Indicates the task assignment success rate; This indicates the number of missions successfully allocated and meeting the constraints of payload capacity, endurance, remaining battery power, mission time window, track continuity, and return-to-home safety. This indicates the total number of tasks to be assigned.

[0080] Flight path planning time refers to the average computation time consumed by the flight path planning algorithm to generate transfer paths and return routes between inspection task points for each UAV after the task allocation results are determined, based on the inspection task point locations, flyable segments, obstacle areas, no-fly zones, turning constraints, energy consumption constraints, and return-to-home safety constraints. This time is the algorithm's computation time and does not include the actual flight time and hovering inspection time of the UAV.

[0081] The scheduling scheme generation time refers to the average computation time consumed from the start of task announcement, candidate UAV screening, bid utility calculation, task allocation, and trajectory coordination optimization at the scheduling center, until an executable multi-UAV collaborative inspection scheduling scheme that meets the aforementioned constraints is output. This metric is used to characterize the computational efficiency of different scheduling methods in generating a complete executable scheme.

[0082] The implementation results show that the method in this embodiment can effectively cover the towers to be inspected, reduce task omissions and duplicate executions, and improve the utilization rate of UAV resources. Simultaneously, by introducing trajectory costs, energy consumption estimation, and return-to-home safety constraints during the task allocation phase, it can reduce flight path redundancy, minimize unnecessary turns and detours, and improve the continuity and executability of flight trajectories. For high-priority tower tasks, the system can prioritize the selection of suitable UAVs and schedule inspections accordingly, thereby improving the response efficiency of power line inspection tasks.

[0083] Example 2: A multi-UAV collaborative task scheduling system for power line inspection, comprising: The input and preprocessing module is configured to acquire multi-source inspection data and preprocess it. The inspection task modeling module is configured to use preprocessed multivariate inspection data to establish a multi-UAV collaborative inspection task scheduling model, objective function, and constraints. The scheduling task output module is configured to generate a scheduling scheme using an improved contract net protocol and an improved ant colony algorithm; The drone swarm management module is used to control drones to take off, inspect, and return according to the scheduling plan.

[0084] In one optional implementation, the input and preprocessing module, the inspection task modeling module, and the scheduling task output module can be deployed in a ground station computer, an edge computing terminal, or a server. The ground station computer, edge computing terminal, or server includes at least a processor, a memory, and a communication interface. The memory stores computer programs for implementing the functions of the corresponding modules. The processor executes these computer programs to acquire and preprocess multi-source inspection data, establish a collaborative inspection task scheduling model, and generate a scheduling scheme.

[0085] The input and preprocessing module can be implemented by a data interface program, a database management program, and a geographic information processing program. It is used to receive basic data of power poles, inspection task requirements, UAV performance status data, operational environment data, and historical inspection task data through the communication interface. It then performs coordinate unification, data cleaning, format conversion, topology relationship establishment, neighbor task set establishment, and candidate flight segment screening on the received data. The data can be stored in a relational database, a spatial database, or a distributed data storage device.

[0086] The inspection task modeling module can be implemented by a task scheduling model building program running in the processor. Based on the preprocessed multivariate inspection data, the task scheduling model building program generates task allocation variables, flight segment selection variables, and associated time and energy variables. It also establishes objective functions and constraints related to task uniqueness, full task coverage, UAV capability matching, endurance, energy consumption, task time window, flight path continuity, takeoff and landing closure, and return safety.

[0087] The scheduling task output module can be implemented by an improved contract network protocol program and an improved ant colony algorithm program running in the processor. The improved contract network protocol program is used to complete candidate UAV screening, task announcement, bid response, bid utility calculation, determination of winning UAV, and local reallocation of conflicting tasks; the improved ant colony algorithm program is used to optimize the task access order, transfer segments between inspection task points, and return paths of each UAV, and form a scheduling scheme including UAV number, assigned task set, task execution order, track node sequence, estimated execution time, and estimated energy consumption.

[0088] The UAV swarm management module can be implemented by a ground station control terminal, communication equipment, and UAV flight controller. The ground station control terminal converts the scheduling scheme into task commands that the corresponding UAVs can recognize through a wired network, wireless LAN, mobile communication network, or dedicated UAV data link, and sends them to the flight controllers of the corresponding UAVs. The flight controllers control the UAVs to perform takeoff, fly to the inspection task point, hover inspection, transfer between inspection task points, and return to home operations in sequence according to the task commands, and feed back the UAV's current position, remaining battery power, flight status, task completion status, and fault status to the ground station control terminal through the communication equipment.

[0089] Each module can operate as an independent software function module, or it can be integrated into the same ground station computer, edge computing terminal, or server; the modules interact with each other through application programming interfaces, message queues, shared databases, or communication networks.

[0090] In one specific software embodiment, the system is deployed using a browser / server architecture. The front-end interactive interface can be implemented using the Vue framework or other web visualization frameworks to display pole locations, drone locations, task allocation results, and flight paths. The back-end can use Python, C++, or Java to write data preprocessing programs, model building programs, improved contract network protocol programs, and improved ant colony algorithm programs. Data storage can use a relational database or a spatial database. Pole locations, obstacles, no-fly zones, and drone flight paths can be displayed using geographic information visualization components. The above software framework is only one implementation method and does not constitute a limitation on the scope of protection of this invention.

[0091] The present invention has been disclosed above with preferred embodiments. However, those skilled in the art should understand that these embodiments are only for describing the present invention and should not be construed as limiting the scope of the present invention. Further improvements can be made without departing from the principles of the present invention, and these improvements should also be considered within the scope of protection of the present invention.

Claims

1. A method for scheduling collaborative tasks involving multiple unmanned aerial vehicles (UAVs) for power line inspection, characterized in that, Includes the following steps: S1. Obtain multi-dimensional inspection data within the area to be inspected and preprocess it. The multi-dimensional inspection data includes power pole foundation data, inspection task requirement data, UAV performance status data, operating environment data, and historical inspection task data. S2. Based on the preprocessed multivariate inspection data, a multi-UAV collaborative inspection task scheduling model is established for collaborative optimization of the task allocation results and flight trajectories of multiple UAVs. The multi-UAV collaborative inspection task scheduling model includes decision variables, objective function and constraints. S3. Based on the improved contract network protocol, before the mission announcement is released, candidate drones are screened according to the aforementioned constraints. During the bidding stage, only candidate drones that meet the requirements of mission capability, expected execution time, expected energy consumption, mission time window, and return safety are allowed to generate bidding responses. The winning drone is determined by combining the bidding utility value. The improved contract network protocol includes: before the mission announcement is released, a candidate drone set is formed based on the drone's payload capacity, inspection payload type, remaining battery power, maximum endurance, current location, and availability, according to the aforementioned constraints; during the bidding phase, only candidate drones whose estimated execution time, estimated energy consumption, mission time window, and return-to-home safety meet the aforementioned constraints are allowed to generate bidding responses; the bidding utility value is calculated based on the mission priority benefit, estimated execution cost, estimated completion time, energy margin, payload matching degree, and current mission load of each candidate drone, and the winning drone is determined based on the bidding utility value; when multiple mission combinations cause the candidate scheduling scheme to fail to meet the constraints of cumulative execution time, cumulative energy consumption, mission time window, track continuity, or return-to-home safety, the relevant tasks are fed back to the mission announcement phase for partial re-announcement, bidding, and evaluation; S4. Based on the improved ant colony algorithm, before each state transition in constructing a candidate scheduling scheme, a set of candidate inspection task points and a set of candidate flight segments are formed according to the constraints, and candidate inspection task points or candidate flight segments that do not meet the constraints are eliminated. The inspection sequence of the UAV, the transfer trajectory between inspection task points and the return path are optimized, and the scheduling scheme is output. The improved ant colony algorithm refers to improvements made to the following parts of the ant colony algorithm: During the ant construction phase, one ant is assigned to one multi-drone collaborative scheduling scheme; In the state transition probability calculation stage, a composite heuristic factor is added, including tower task priority, node distance, expected time cost, expected energy consumption, and constraint feasibility. In the next inspection task point selection phase, an initial candidate inspection task point set is formed based on the topological adjacency relationship, the set of neighboring tasks, and the set of candidate flight segments corresponding to the current inspection task point, and the initial candidate inspection task point set is filtered according to the constraints. The state transition probability is calculated only for candidate inspection task points that meet the constraints; and the tower task priority, node distance, spatial proximity, topological continuity, estimated time cost and estimated energy consumption are used together as composite heuristic information to determine the next inspection task point of the UAV, and the state transition probability is calculated only for the inspection task points in the set of candidate inspection task points. When the candidate inspection task point set is empty, or when multiple task combinations generate cumulative constraint conflicts, the relevant tasks will be fed back to the improved contract network protocol for local reallocation.

2. The method for scheduling collaborative tasks of multiple unmanned aerial vehicles (UAVs) for power line inspection according to claim 1, characterized in that, The basic data for the power poles includes the pole number to be inspected, two-dimensional horizontal coordinates, pole elevation spatial coordinates, task height, line affiliation, equipment type, connection relationship between adjacent poles, and spatial distribution relationship; the pole elevation serves as the task height attribute for the corresponding inspection task; the inspection task requirement data includes inspection type, historical fault frequency, equipment criticality, inspection urgency, task time window, and task completion requirements; the drone performance status data includes the number of drones, current location, maximum endurance, flight speed, turn rate, minimum turning radius, payload capacity, remaining battery power, availability status, and historical data. The system tracks the completion status of historical inspection tasks. If the drone is in normal online operation, without malfunctions or communication abnormalities, not under maintenance, and not occupied by other tasks conflicting with the target inspection task, it is determined to be in an available state; otherwise, it is determined to be in an unavailable state. The operational environment data includes wind speed, wind direction, terrain environment, obstacle information, and no-fly zone information. The historical inspection task data includes historical task type, historical task priority, historical task execution time, historical task energy consumption, historical task completion status, historical fault or anomaly records, and wind speed and direction environmental parameters during task execution.

3. The method for scheduling collaborative tasks of multiple unmanned aerial vehicles (UAVs) for power line inspection according to claim 1, characterized in that, The preprocessing includes: mapping the horizontal position of each tower to be inspected to a two-dimensional plane coordinate system; establishing a topological adjacency matrix based on the connection relationship between adjacent towers; establishing a neighboring task set based on the two-dimensional horizontal coordinates and spatial distribution relationship of each tower to be inspected; retaining the task height of each tower to be inspected as a vertical attribute associated with the corresponding two-dimensional horizontal coordinate; and determining the reachability of the flight paths between inspection task points based on obstacle information and no-fly zone information, establishing a candidate flight segment set, and simultaneously establishing a set of towers to be inspected and a set of drones.

4. The method for scheduling collaborative tasks of multiple unmanned aerial vehicles (UAVs) for power line inspection according to claim 1, characterized in that, The objective function includes: a task allocation completeness index, used to evaluate whether all poles to be inspected have been uniquely assigned tasks; a UAV resource utilization rate index, used to evaluate the resource utilization of UAVs participating in the inspection tasks; a comprehensive trajectory cost index, used to evaluate trajectory costs based on the flight distance and energy consumption generated by UAVs during the execution of inspection tasks; a high-priority task delay penalty index, used to evaluate the scheduling cost caused by the failure to complete high-priority inspection tasks in a timely manner; and a task load balancing index, used to evaluate the balance between the number of inspection tasks undertaken by each UAV or the workload of the tasks. The objective function is a weighted sum of the task allocation completeness index, the UAV resource utilization rate index, the comprehensive trajectory cost, and the priority penalty term.

5. A method for scheduling collaborative tasks involving multiple unmanned aerial vehicles (UAVs) for power line inspection according to claim 1, characterized in that, The constraints include mission uniqueness constraint, mission full coverage constraint, UAV capability matching constraint, maximum endurance constraint, energy consumption constraint, mission time window constraint, flight path continuity constraint, takeoff and landing closure constraint, and return safety constraint.

6. The method for scheduling collaborative tasks of multiple unmanned aerial vehicles (UAVs) for power line inspection according to claim 1, characterized in that, S4 further includes the following steps: when the generated scheduling scheme does not meet at least one constraint, the corresponding task is fed back to the improved contract network protocol for local reallocation, and a new scheduling scheme is generated.

7. A method for scheduling collaborative tasks involving multiple unmanned aerial vehicles (UAVs) for power line inspection according to claim 1, characterized in that, It also includes the following steps: During the inspection mission of the UAV, when an unexpected event occurs that affects the UAV inspection, the constraints are regenerated, and a new scheduling scheme is generated by reusing the improved contract net protocol and the improved ant colony algorithm. The unforeseen events affecting drone inspections include new inspection tasks, wind speed, wind direction, drone malfunction, and obstruction of at least one drone flight path.

8. A multi-UAV collaborative task scheduling system for power line inspection, characterized in that, include: The input and preprocessing module is configured to acquire multi-source inspection data and preprocess it. The inspection task modeling module is configured to use preprocessed multivariate inspection data to establish a multi-UAV collaborative inspection task scheduling model, objective function, and constraints. The scheduling task output module is configured to generate a scheduling scheme using an improved contract net protocol and an improved ant colony algorithm; The drone swarm management module is configured to distribute the scheduling scheme to the corresponding drones and control multiple drones to perform takeoff, inspection, and return according to the scheduling scheme; wherein, The improved contract network protocol includes: before the mission announcement is released, a candidate drone set is formed based on the drone's payload capacity, inspection payload type, remaining battery power, maximum endurance, current location, and availability, according to the aforementioned constraints; during the bidding phase, only candidate drones whose estimated execution time, estimated energy consumption, mission time window, and return-to-home safety meet the aforementioned constraints are allowed to generate bidding responses; the bidding utility value is calculated based on the mission priority benefit, estimated execution cost, estimated completion time, energy margin, payload matching degree, and current mission load of each candidate drone, and the winning drone is determined based on the bidding utility value; when multiple mission combinations cause the candidate scheduling scheme to fail to meet the constraints of cumulative execution time, cumulative energy consumption, mission time window, track continuity, or return-to-home safety, the relevant tasks are fed back to the mission announcement phase for partial re-announcement, bidding, and evaluation; The improved ant colony algorithm refers to improvements made to the following parts of the ant colony algorithm: During the ant construction phase, one ant is assigned to one multi-drone collaborative scheduling scheme; In the state transition probability calculation stage, a composite heuristic factor is added, including tower task priority, node distance, expected time cost, expected energy consumption, and constraint feasibility. In the next inspection task point selection phase, an initial candidate inspection task point set is formed based on the topological adjacency relationship, the set of neighboring tasks, and the set of candidate flight segments corresponding to the current inspection task point, and the initial candidate inspection task point set is filtered according to the constraints. The state transition probability is calculated only for candidate inspection task points that meet the constraints; and the tower task priority, node distance, spatial proximity, topological continuity, estimated time cost and estimated energy consumption are used together as composite heuristic information to determine the next inspection task point of the UAV, and the state transition probability is calculated only for the inspection task points in the set of candidate inspection task points. When the candidate inspection task point set is empty, or when multiple task combinations generate cumulative constraint conflicts, the relevant tasks will be fed back to the improved contract network protocol for local reallocation.

Images