Multi-unmanned aerial vehicle dynamic task allocation method and device, medium and equipment

Abstract

The invention discloses a multi-unmanned aerial vehicle dynamic task allocation method and device, a medium and equipment. The method comprises the steps of obtaining an inspection task and an initial state parameter of each unmanned aerial vehicle; based on the task units and the initial state parameters of each unmanned aerial vehicle, performing preliminary alliance division, and performing combination iteration on the multiple preliminarily divided alliances to obtain an optimal alliance; in the inspection process of each unmanned aerial vehicle according to the optimal alliance strategy, the real-time state parameter of each unmanned aerial vehicle is obtained, the alliance state parameter is generated based on the real-time state parameter of each unmanned aerial vehicle, and whether an emergency occurs or not is judged based on the real-time state parameter of each unmanned aerial vehicle and the newly added task priority; and when an emergency occurs, real-time state data is generated based on the emergency, the real-time state parameters of each unmanned aerial vehicle and the alliance state parameters, and the real-time state data is input to the trained strategy adjustment model to obtain an adjusted task strategy, so that the strategy is rapidly and accurately adjusted, and the overall inspection efficiency of the unmanned aerial vehicle cluster is improved.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) task allocation technology, and in particular to a method, apparatus, medium and equipment for dynamic task allocation of multiple UAVs. Background Technology

[0002] With the transformation of my country's energy structure and the advancement of dual-carbon goals, the western region, with its abundant solar and wind energy resources, has become an important clean energy base for the country. The concentrated construction of large-scale photovoltaic power plants and wind farms has made the western power grid play a crucial role in the West-to-East Power Transmission strategy. To ensure the safe and stable operation of long-distance, high-load transmission lines, regular, efficient, and reliable power line inspections are essential.

[0003] Given the complex terrain traversed by power transmission lines in western China, encompassing diverse landforms such as plateaus, mountains, and deserts, traditional manual inspections or single-drone inspections suffer from low efficiency, limited coverage, and high personnel safety risks. Therefore, drone swarm inspection technology has become the mainstream choice. Currently, in drone swarm inspections, the multi-drone task allocation process mainly relies on human experience to formulate strategies. That is, staff manually divide the inspection task area for each drone based on experience values ​​such as the length of the line section and the number of devices.

[0004] However, the existing drone swarm task allocation method has obvious technical defects: on the one hand, the reliance on human experience leads to uneven task allocation, with some drones bearing an excessively heavy inspection load (such as covering long-distance complex lines), while some drones are under light load, resulting in low overall swarm inspection efficiency; on the other hand, when faced with sudden failures (such as drone equipment failure or sudden weather changes in the inspection area), the existing system lacks a rapid response mechanism, and the response speed of load balancing adjustment and real-time task reallocation is slow, which easily leads to interruption, omission, or repeated inspection of inspection tasks, seriously affecting the timeliness and completeness of power transmission line inspections in western regions, and making it difficult to meet the high-efficiency requirements of large-scale clean energy bases for power grid safety operation and maintenance. Summary of the Invention

[0005] In view of this, the present invention provides a method, apparatus, medium and equipment for dynamic task allocation of multiple UAVs, the main purpose of which is to solve the problem of low overall inspection efficiency of the cluster caused by existing dynamic task allocation methods for multiple UAVs.

[0006] According to one aspect of this application, a method for dynamic task allocation among multiple unmanned aerial vehicles (UAVs) is provided, the method comprising: The inspection task of power transmission lines is divided into multiple task units, and the initial state parameters of each drone are obtained. Based on the initial state parameters of each task unit and each UAV, the alliance is initially divided to obtain multiple initial alliances. The multiple initial alliances are merged and iterated to obtain the optimal alliance. Based on the optimal alliance, the initial task strategy of each UAV is determined. During the inspection process of each UAV according to the initial mission strategy, the real-time status parameters of each UAV are obtained, the alliance status parameters are generated based on the real-time status parameters of each UAV, the newly added mission priority is obtained, and the occurrence of a sudden event is determined based on the real-time status parameters of each UAV and the newly added mission priority. When an emergency occurs, real-time status data is generated based on the emergency, the real-time status parameters of each UAV, and the status parameters of the alliance. The real-time status data is then input into the trained strategy adjustment model to obtain the adjusted task strategy.

[0007] Optionally, based on the initial state parameters of each task unit and each UAV, the initial division of the alliance is performed to obtain multiple initial alliances, including: Based on the task position information of the task unit and the initial position information of the drone, the straight-line distance between each task unit and each drone is calculated, and the idle drone with the closest straight-line distance to the task unit is taken as the single alliance of the drone. Determine whether each single alliance satisfies the boundary constraints. If there is a single alliance that does not satisfy the boundary constraints, reassign the drones corresponding to the single alliance that does not satisfy the boundary constraints and its neighboring drones to the task units until each single alliance satisfies the boundary constraints, thus obtaining multiple initial alliances.

[0008] Optionally, the step of merging and iterating the multiple initial alliances to obtain the optimal alliance includes: Calculate the total travel time and total mission time of each drone within the alliance, obtain the initial available time of each drone, and calculate the revenue of a single alliance based on the total travel time, total mission time and initial available time of each drone. Merge any two initial alliances, calculate the total revenue of the merged alliance, and calculate the merged revenue increment based on the total revenue and the revenue of each individual alliance. If the merged revenue increment is greater than zero, the merged alliance is used as the new initial alliance; if the merged revenue increment is less than zero, the original initial alliance is retained. Repeat the above process until there are no two initial alliances whose combined profit increment is greater than zero, thus obtaining the optimal initial alliance.

[0009] Optionally, the revenue of a single consortium can be calculated using the following method.

[0010]

[0011]

[0012]

[0013] Where E(S) is the efficiency gain, Q(S) is the quality gain, and B(S) is the equilibrium gain. , , For weights, d j For the drones in the initial alliance j Total movement time within the alliance, c i To complete the mission for drones within the initial alliance i Task time e j This is the initial available time for all drones within the initial alliance; As task priority weight, This is the equipment efficiency coefficient. This represents the average load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance.

[0014] Optionally, the step of determining whether a sudden event has occurred based on the real-time status parameters of each of the UAVs and the newly added task priority includes: The fault markers of the UAV are obtained from the real-time status parameters of the UAV. Based on the newly added task priority, task units with a new task priority level greater than a preset level are identified. When the fault marker indicates that a fault has occurred and / or the newly added task priority level is greater than the preset level, an emergency event is determined to have occurred.

[0015] Optionally, the alliance status parameters include alliance task completion rate, total task time within the alliance, alliance load difference rate, and proportion of high-priority tasks within the alliance. Generating alliance status parameters based on the real-time status parameters of each UAV includes: The alliance mission completion rate is calculated based on the number of missions completed by drones within the alliance and the total number of missions included in the alliance. The total time for the alliance mission is the sum of the time taken by each drone to complete its mission and the time taken to travel. The load rate of a single drone is calculated based on the total mission time of each drone as a percentage of its initial available time. The average load rate of all drones in the alliance is calculated based on the load rates of all single drones. The ratio of the difference between the maximum and minimum load rates of all drones in the alliance to the average load rate of all drones is used as the alliance load difference rate. Calculate the percentage of high-priority tasks within the alliance based on the priority of task units within the alliance.

[0016] Optionally, the trained policy adjustment model can be obtained using the following methods: Acquire historical mission information of power transmission lines, mission strategies of drones and historical real-time status data of drones during historical inspections, records of emergencies, manual adjustment strategies and the effects of adjustment strategies; Historical alliance status data is calculated based on the historical real-time status data. Historical status data is generated by combining the historical alliance status data, historical real-time status data, and emergency event records. Historical action data is generated based on the manual adjustment strategy corresponding to the emergency record. Historical rewards corresponding to the historical action data are calculated based on the effect of the adjustment strategy and the preset reward rules. Based on the historical state data, historical action data, and historical rewards, the initial policy adjustment model is trained to obtain the trained policy adjustment model.

[0017] According to another aspect of this application, a multi-UAV dynamic task allocation device is provided, comprising: The initial data acquisition module is used to divide the inspection task of the power transmission line into multiple task units and acquire the initial state parameters of each drone. The initial mission strategy determination module is used to perform preliminary division of the alliance based on the initial state parameters of each mission unit and each UAV, obtain multiple initial alliances, merge and iterate the multiple initial alliances to obtain the optimal alliance, and determine the initial mission strategy of each UAV based on the optimal alliance. The real-time data acquisition module is used to acquire the real-time status parameters of each UAV during the inspection process of each UAV according to the initial task strategy, generate alliance status parameters based on the real-time status parameters of each UAV, acquire the newly added task priority, and determine whether an emergency has occurred based on the real-time status parameters of each UAV and the newly added task priority. The task strategy adjustment module is used to generate real-time status data based on the sudden event, the real-time status parameters of each UAV, and the alliance status parameters when an emergency occurs. The real-time status data is then input into the trained strategy adjustment model to obtain the adjusted task strategy.

[0018] Optionally, the initial task strategy determination module is further configured to: Based on the task position information of the task unit and the initial position information of the drone, the straight-line distance between each task unit and each drone is calculated, and the idle drone with the closest straight-line distance to the task unit is taken as the single alliance of the drone. Determine whether each single alliance satisfies the boundary constraints. If there is a single alliance that does not satisfy the boundary constraints, reassign the drones corresponding to the single alliance that does not satisfy the boundary constraints and its neighboring drones to the task units until each single alliance satisfies the boundary constraints, thus obtaining multiple initial alliances.

[0019] Optionally, the initial task strategy determination module is further configured to: Calculate the total travel time and total mission time of each drone within the alliance, obtain the initial available time of each drone, and calculate the revenue of a single alliance based on the total travel time, total mission time and initial available time of each drone. Merge any two initial alliances, calculate the total revenue of the merged alliance, and calculate the merged revenue increment based on the total revenue and the revenue of each individual alliance. If the merged revenue increment is greater than zero, the merged alliance is used as the new initial alliance; if the merged revenue increment is less than zero, the original initial alliance is retained. Repeat the above process until there are no two initial alliances whose combined profit increment is greater than zero, thus obtaining the optimal initial alliance.

[0020] Optionally, the revenue of a single consortium can be calculated using the following method.

[0021]

[0022]

[0023]

[0024] Where E(S) is the efficiency gain, Q(S) is the quality gain, and B(S) is the equilibrium gain. , , For weights, d j For the drones in the initial alliance j Total movement time within the alliance, c i To complete the mission for drones within the initial alliance i Task time e j This is the initial available time for all drones within the initial alliance; As task priority weight, This is the equipment efficiency coefficient. This represents the average load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance.

[0025] Optionally, the initial task strategy determination module is further configured to: The fault markers of the UAV are obtained from the real-time status parameters of the UAV. Based on the newly added task priority, task units with a new task priority level greater than a preset level are identified. When the fault marker indicates that a fault has occurred and / or the newly added task priority level is greater than the preset level, an emergency event is determined to have occurred.

[0026] Optionally, the initial task strategy determination module is further configured to: The alliance mission completion rate is calculated based on the number of missions completed by drones within the alliance and the total number of missions included in the alliance. The total time for the alliance mission is the sum of the time taken by each drone to complete its mission and the time taken to travel. The load rate of a single drone is calculated based on the total mission time of each drone as a percentage of its initial available time. The average load rate of all drones in the alliance is calculated based on the load rates of all single drones. The ratio of the difference between the maximum and minimum load rates of all drones in the alliance to the average load rate of all drones is used as the alliance load difference rate. Calculate the percentage of high-priority tasks within the alliance based on the priority of task units within the alliance.

[0027] Optionally, the task strategy adjustment module is also used for: Acquire historical mission information of power transmission lines, mission strategies of drones and historical real-time status data of drones during historical inspections, records of emergencies, manual adjustment strategies and the effects of adjustment strategies; Historical alliance status data is calculated based on the historical real-time status data. Historical status data is generated by combining the historical alliance status data, historical real-time status data, and emergency event records. Historical action data is generated based on the manual adjustment strategy corresponding to the emergency record. Historical rewards corresponding to the historical action data are calculated based on the effect of the adjustment strategy and the preset reward rules. Based on the historical state data, historical action data, and historical rewards, the initial policy adjustment model is trained to obtain the trained policy adjustment model.

[0028] According to another aspect of this application, a storage medium is provided that stores at least one executable instruction, which causes a processor to perform the operation corresponding to the multi-UAV dynamic task allocation method described above.

[0029] According to another aspect of this application, a computer device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the above-described multi-UAV dynamic task allocation method.

[0030] By employing the above-described technical solutions, the technical solutions provided by the embodiments of the present invention have at least the following advantages: This application provides a method, apparatus, device, and medium for dynamic task allocation of multiple drones. Based on game theory, it establishes an optimal alliance strategy for drone inspection, obtains an initial task strategy that balances benefits and costs, acquires real-time state parameters of drones during inspection according to the initial task strategy, generates alliance state parameters based on the real-time state parameters of drones, and automatically identifies sudden events by combining the real-time state parameters of drones and newly added task priorities. When a sudden event occurs, it generates real-time state data by combining the sudden event, the real-time state parameters of each drone, and the alliance state parameters. The real-time state data is input into a trained strategy adjustment model to obtain an adjusted task strategy. This achieves rapid strategy adjustment without manual intervention, thereby improving the overall inspection efficiency of the drone swarm.

[0031] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0032] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart of a multi-UAV dynamic task allocation method provided in an embodiment of this application is shown; Figure 2 This illustration shows another flowchart of a multi-UAV dynamic task allocation method provided in an embodiment of this application; Figure 3This paper shows a block diagram of a multi-UAV dynamic task allocation device provided in an embodiment of this application; Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention is shown.

[0033] in, Figure 3 In the middle section: 302 - Initial data acquisition module; 304 - Initial task strategy determination module; 306 - Real-time data acquisition module; 308 - Task strategy adjustment module; Figure 4 In Chinese: 402 - Processor; 404 - Communication interface; 406 - Memory; 408 - Communication bus; 410 - Program. Detailed Implementation

[0034] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present invention can be combined with each other.

[0035] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the specific embodiments, structures, features, and effects according to the present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "embodiments" or "embodiments" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0036] To address the low accuracy of current multi-UAV dynamic task allocation methods, this application provides a multi-UAV dynamic task allocation method, such as... Figure 1 As shown, the method includes: 102: Divide the inspection task of the power transmission line into multiple task units and obtain the initial state parameters of each drone; 104: Based on the initial state parameters of each task unit and each UAV, the alliance is initially divided to obtain multiple initial alliances. The multiple initial alliances are merged and iterated to obtain the optimal alliance. Based on the optimal alliance, the initial task strategy of each UAV is determined. 106: During the inspection process of each drone according to the initial mission strategy, the real-time status parameters of each drone are obtained, the alliance status parameters are generated based on the real-time status parameters of each drone, the newly added mission priority is obtained, and the sudden event is determined based on the real-time status parameters of each drone and the newly added mission priority. 108: When an emergency occurs, real-time status data is generated based on the emergency, the real-time status parameters of each UAV, and the alliance status parameters. The real-time status data is then input into the trained strategy adjustment model to obtain the adjusted task strategy.

[0037] In this embodiment, the inspection task of the power transmission line is divided into multiple task units. The parameters of each task unit and the initial state parameters of each UAV are obtained. The parameters of the task unit include task type, task location, task attributes, etc., and the initial state parameters of the UAV include UAV position, initial battery level, maximum flight speed, and maximum control radius, etc.

[0038] Based on the parameters of the task unit and the initial state parameters of the UAV, the task unit is preferentially assigned to the UAV with "geographical proximity and capability compatibility" to form a basic single alliance (1 task unit + 1 UAV). This avoids the formation of single alliances from ineffective cross-regional flights, resulting in multiple initial alliances. To improve inspection efficiency, the task units are merged, that is, the initial alliances are merged and optimized. After multiple iterations of merging and optimization, the optimal alliance is obtained, and the UAV uses the task allocation strategy of the optimal alliance as its initial task strategy.

[0039] During the inspection process of each drone according to the initial mission strategy, real-time status parameters of the drones are acquired. The real-time status parameters of individual drones are aggregated according to the alliance hierarchy, and then calculated to obtain alliance status data reflecting the overall collaborative status of the alliance, such as alliance mission completion rate, alliance load difference rate, and alliance remaining available time. Two types of emergencies are used as trigger conditions, and the condition of either one is considered an emergency: one is equipment failure or remaining battery power less than a preset value, and the other is a new task with a priority higher than a preset level or an existing task with a priority higher than a preset level. In this case, in order to reduce the impact of emergencies, real-time status data is generated from the emergencies, the real-time status parameters of each drone, and the alliance status parameters. The real-time status data is then input into a trained strategy adjustment model to obtain the adjusted mission strategy.

[0040] This application provides a method for dynamic task allocation among multiple drones. Compared with existing technologies, it is based on game theory to establish an optimal alliance strategy for drone inspection, thereby obtaining an initial task strategy that balances benefits and costs. It acquires the real-time status parameters of the drones during the inspection process according to the initial task strategy, generates alliance status parameters based on the real-time status parameters of the drones, and automatically identifies sudden events by combining the real-time status parameters of the drones and the newly added task priorities. When a sudden event occurs, it generates real-time status data by combining the sudden event, the real-time status parameters of each drone, and the alliance status parameters. The real-time status data is then input into a trained strategy adjustment model to obtain the adjusted task strategy. This method achieves rapid strategy adjustment without manual intervention, thereby improving the overall inspection efficiency of the drone swarm.

[0041] In one embodiment, such as Figure 2 As shown, based on the initial state parameters of each task unit and each UAV, the alliance is initially divided into multiple initial alliances. These initial alliances are then merged and iterated to obtain the optimal alliance, which includes: 202: Based on the task position information of the task unit and the initial position information of the drone, calculate the straight-line distance between each task unit and each drone, and select the idle drone with the closest straight-line distance to the task unit as the drone alliance. 204: Determine whether each single alliance satisfies the boundary constraints. If there is a single alliance that does not satisfy the boundary constraints, the drones corresponding to the single alliance that does not satisfy the boundary constraints and its neighboring drones are reassigned to the task units until each single alliance satisfies the boundary constraints, thus obtaining multiple initial alliances. 206: Calculate the total travel time and total mission time of each drone within the alliance, obtain the initial available time of each drone, and calculate the revenue of a single alliance based on the total travel time, total mission time and initial available time of each drone. 208: Merge any two initial alliances, calculate the total revenue of the merged alliance, and calculate the incremental revenue of the merged alliance based on the total revenue and the revenue of each individual alliance. If the incremental revenue of the merged alliance is greater than zero, the merged alliance is used as the new initial alliance. If the incremental revenue of the merged alliance is less than zero, the original initial alliance is retained. 210: Repeat the above process until there are no two initial alliances whose combined profit increment is greater than zero, and obtain the optimal initial alliance.

[0042] Specifically, drones that are not currently assigned any task units, are in normal condition, and have sufficient remaining power to meet the minimum execution requirements are considered idle drones. Based on the geographical coordinates of the task unit and the initial position coordinates of the drone, the straight-line distance between the task unit and the drone is calculated using the Euclidean distance formula. The drones with the closest straight-line distance to the task unit are then paired with the task unit to form a single alliance.

[0043] The boundary constraints include: (1) each task unit belongs to only one alliance; (2) the total task time of the alliance to which a single UAV belongs is less than or equal to its initial available time; (3) task units within an alliance need to be geographically close to each other to avoid wasting energy by flying UAVs across regions. Identify single alliances that do not meet the boundary constraints, and reallocate the unsatisfied task units to UAVs until all boundary constraints are met, thus obtaining multiple initial alliances.

[0044] For each single alliance, the benefit, quality, and equilibrium benefits of the single alliance are calculated based on the task time taken by the drone to complete the task, the initial available time of the drone, the equipment efficiency coefficient, and the task priority weight. The weights of the benefit, quality, and equilibrium benefits are assigned according to their importance, and the benefit of the single alliance is calculated based on the weights and the benefit, quality, and equilibrium benefits.

[0045] The process involves merging two initial alliances. Based on factors such as drone power consumption or high-definition capabilities, tasks within the merged alliance are assigned to suitable drones. Task execution parameters are obtained from the drones within the alliance based on their projected performance. The total revenue of the merged alliance is calculated, and the individual revenues of the two single alliances are subtracted from the total revenue to obtain the merged revenue increment. If the merged revenue increment is greater than zero, the two single alliances are suitable for merging, and the merged alliance becomes the new initial alliance. If the merged revenue increment is less than zero, the two single alliances are not suitable for merging, and they are retained. The process of merging these two single alliances with other single alliances is then repeated. Each iteration iterates through all existing initial alliance pairs to ensure no merging possibilities are missed. This "merge-calculate increment-judgment" process is repeated until the revenue increment after merging all possible alliance pairs is less than zero. At this point, the alliance structure is optimal, meaning that merging cannot yield higher revenue.

[0046] The final output of the optimal alliance enables multiple drones to cooperate more efficiently and achieve better results during inspections, providing a high-quality initial strategy for subsequent dynamic task adjustments.

[0047] In this embodiment, the revenue of a single consortium is calculated using the following method.

[0048]

[0049]

[0050]

[0051] Where E(S) is the efficiency gain, Q(S) is the quality gain, and B(S) is the equilibrium gain. , , For weights, d j For the drones in the initial alliance j Total movement time within the alliance, c i To complete the mission for drones within the initial alliance i Task time e j This is the initial available time for all drones within the initial alliance; As task priority weight, This is the equipment efficiency coefficient. This represents the average load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance.

[0052] Specifically, the parameter information for each task unit includes priority weight, location information, and standard time. The initial state parameters for the drone include initial available time, equipment efficiency coefficient, and initial position coordinates. Real-time status data of the drone during task execution is acquired, such as task completion time, movement time, and power consumption. Efficiency gain is the ratio of the total time the alliance takes to complete the task to the total available time of the drones; a smaller ratio is better. Quality gain is the weighted sum of task completion quality within the alliance (equipment efficiency × task priority). Balanced gain is the drone load difference rate within the alliance (a smaller difference is better). The total gain is calculated by weighting efficiency gain, quality gain, and balanced gain according to their importance.

[0053] In one embodiment, determining whether a sudden event has occurred based on the real-time status parameters of each drone and the newly added task priority includes: The fault markers of the UAV are obtained from the real-time status parameters of the UAV. Based on the newly added task priority, the task units with a new task priority level greater than the preset level are identified. When the fault marker indicates that a fault has occurred and / or the newly added task priority level is greater than the preset level, it is determined that an emergency has occurred.

[0054] In one embodiment, the alliance status parameters include alliance task completion rate, total task time within the alliance, alliance load difference rate, and the proportion of high-priority tasks within the alliance. These alliance status parameters are generated based on the real-time status parameters of each drone and include: The alliance mission completion rate is calculated based on the number of missions completed by drones within the alliance and the total number of missions included in the alliance. The total time for the alliance mission is the sum of the time taken by each drone to complete its mission and the time taken to travel. The load rate of a single drone is calculated based on the total mission time of each drone as a percentage of its initial available time. The average load rate of all drones in the alliance is calculated based on the load rates of all single drones. The ratio of the difference between the maximum and minimum load rates of all drones in the alliance to the average load rate of all drones is used as the alliance load difference rate. Calculate the percentage of high-priority tasks within the alliance based on the priority of task units within the alliance.

[0055] Specifically, the total time for the alliance's missions is calculated based on the time taken by each drone to complete its mission and its travel time. The execution order of the tasks that constitute the total mission time will change the travel distance (e.g., a "near → far" order is shorter than a "far → near" order). Therefore, before calculating the travel time, tasks must be sorted by "geographical proximity + priority" (higher priority tasks take precedence, and tasks of the same priority are sorted by "near → far") to ensure that the time calculation closely matches the actual execution path. If multiple drones in the alliance are collaboratively executing the same batch of tasks, the travel time must be calculated separately for each drone's independent path and then summed.

[0056] Calculate the load rate of each drone, which is the ratio of the total time spent on the current task to its initial available time. Calculate the average load rate of all drones in the alliance based on the load rates of all individual drones. Calculate the alliance load difference rate based on the maximum, minimum, and average load rates of all drones in the alliance.

[0057] Alliance status parameters reflect the overall task progress, coordination efficiency, and load balance of the alliance, providing a basis for decision-making regarding alliance splitting / merging, task transfer, and other actions.

[0058] In one embodiment, the trained policy adjustment model is obtained using the following method: Acquire historical mission information of power transmission lines, mission strategies of drones and historical real-time status data of drones during historical inspections, records of emergencies, manual adjustment strategies and the effects of adjustment strategies; Historical alliance status data is calculated based on historical real-time status data. Historical status data is generated by combining historical alliance status data, historical real-time status data, and emergency event records. Historical action data is generated based on the manual adjustment strategies corresponding to the emergency records. Historical rewards corresponding to the historical action data are calculated based on the effects of the adjustment strategies and the preset reward rules. Based on historical state data, historical action data, and historical rewards, the initial policy adjustment model is trained to obtain the trained policy adjustment model.

[0059] Specifically, this involves acquiring all the basic information required for training, such as historical inspection mission information (e.g., inspection route segment, mission objective, planned duration), the UAV's initial mission strategy (e.g., flight path, inspection speed, sensor parameters), historical real-time status data (e.g., flight position, battery level, sensor accuracy, flight attitude), records of unexpected events (e.g., line icing, UAV malfunction, severe weather), manual adjustment strategies (e.g., flight path modifications and pause / restart commands after human detection of anomalies), and the effects of manual adjustment strategies (e.g., whether the inspection was successfully completed after adjustment, whether the anomaly was resolved, and changes in mission time).

[0060] First, historical alliance status data is calculated based on historical real-time status data. Then, the alliance status data, UAV historical real-time status data, and incident records are merged to form historical status data describing a patrol scenario at a specific moment. For each incident record, the manual adjustment strategy made by humans at that time is extracted as the action in that scenario. Based on the effect of the manual adjustment strategy (e.g., a successful return without loss is a good effect, and a UAV losing contact is a bad effect), combined with preset reward rules (e.g., +10 points for a good effect and -20 points for a bad effect), each action is scored. The processed historical status data, historical action data, and historical reward data are input into an initial strategy adjustment model (e.g., a reinforcement learning model). Through repeated learning and optimization, the model will continuously try to choose which action to obtain higher rewards in a certain state, gradually mimicking the logic of humans handling patrol anomalies.

[0061] When the model's decision-making performance (such as the inspection success rate and efficiency after adjusting the strategy) stabilizes and meets the target, the training is complete, and a model that can autonomously adjust the task strategy according to the real-time inspection status is finally obtained.

[0062] Furthermore, as a response to the above Figure 1 The implementation of the method shown in this invention provides a multi-UAV dynamic task allocation device, such as... Figure 3 As shown, the device includes: The initial data acquisition module 302 is used to divide the inspection task of the power transmission line into multiple task units and acquire the initial state parameters of each UAV. The initial mission strategy determination module 304 is used to perform preliminary division of the alliance based on the initial state parameters of each mission unit and each UAV, obtain multiple initial alliances, merge and iterate the multiple initial alliances to obtain the optimal alliance, and determine the initial mission strategy of each UAV based on the optimal alliance. The real-time data acquisition module 306 is used to acquire the real-time status parameters of each UAV during the inspection process of each UAV according to the initial mission strategy, generate alliance status parameters based on the real-time status parameters of each UAV, acquire the newly added mission priority, and determine whether an emergency has occurred based on the real-time status parameters of each UAV and the newly added mission priority. The task strategy adjustment module 308 is used to generate real-time status data based on the sudden event, the real-time status parameters of each UAV, and the alliance status parameters when a sudden event occurs. The real-time status data is then input into the trained strategy adjustment model to obtain the adjusted task strategy.

[0063] This application provides a multi-UAV dynamic task allocation device. Compared with the prior art, it is based on game theory to form an optimal alliance strategy for UAV inspection, obtain an initial task strategy that takes into account both benefits and costs, obtain the real-time status parameters of UAVs during the inspection process according to the initial task strategy, generate alliance status parameters based on the real-time status parameters of UAVs, and automatically identify sudden events by combining the real-time status parameters of UAVs and the newly added task priorities. When a sudden event occurs, the sudden event, the real-time status parameters of each UAV, and the alliance status parameters are used to generate real-time status data. The real-time status data is input into a trained strategy adjustment model to obtain the adjusted task strategy. This achieves rapid strategy adjustment without manual intervention, thereby improving the overall inspection efficiency of the UAV swarm.

[0064] In one embodiment, the initial task policy determination module is further configured to: Based on the task position information of the task unit and the initial position information of the drone, calculate the straight-line distance between each task unit and each drone, and select the idle drone with the closest straight-line distance to the task unit as the drone alliance. Determine whether each single alliance satisfies the boundary constraints. If there is a single alliance that does not satisfy the boundary constraints, reassign the drones corresponding to the single alliance that does not satisfy the boundary constraints and its neighboring drones to the task units until each single alliance satisfies the boundary constraints, thus obtaining multiple initial alliances.

[0065] In one embodiment, the initial task policy determination module is further configured to: Calculate the total travel time and total mission time of each drone within the alliance, obtain the initial available time of each drone, and calculate the revenue of a single alliance based on the total travel time, total mission time and initial available time of each drone. Merge any two initial alliances, calculate the total revenue of the merged alliance, and calculate the incremental revenue of the merged alliance based on the total revenue and the revenue of each individual alliance. If the incremental revenue of the merged alliance is greater than zero, the merged alliance is used as the new initial alliance. If the incremental revenue of the merged alliance is less than zero, the original initial alliance is retained. Repeat the above process until there are no two initial alliances whose combined profit increment is greater than zero, thus obtaining the optimal initial alliance.

[0066] In one embodiment, the revenue of a single consortium is calculated using the following method.

[0067]

[0068]

[0069]

[0070] Where E(S) is the efficiency gain, Q(S) is the quality gain, and B(S) is the equilibrium gain. , , For weights, d j For the drones in the initial alliance j Total movement time within the alliance, c i To complete the mission for drones within the initial alliance i Task time e j This is the initial available time for all drones within the initial alliance; As task priority weight, This is the equipment efficiency coefficient. This represents the average load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance.

[0071] In one embodiment, the initial task policy determination module is further configured to: The fault markers of the UAV are obtained from the real-time status parameters of the UAV. Based on the newly added task priority, the task units with a new task priority level greater than the preset level are identified. When the fault marker indicates that a fault has occurred and / or the newly added task priority level is greater than the preset level, it is determined that an emergency has occurred.

[0072] In one embodiment, the initial task policy determination module is further configured to: The alliance mission completion rate is calculated based on the number of missions completed by drones within the alliance and the total number of missions included in the alliance. The total time for the alliance mission is the sum of the time taken by each drone to complete its mission and the time taken to travel. The load rate of a single drone is calculated based on the total mission time of each drone as a percentage of its initial available time. The average load rate of all drones in the alliance is calculated based on the load rates of all single drones. The ratio of the difference between the maximum and minimum load rates of all drones in the alliance to the average load rate of all drones is used as the alliance load difference rate. Calculate the percentage of high-priority tasks within the alliance based on the priority of task units within the alliance.

[0073] In one embodiment, the task strategy adjustment module is further configured to: Acquire historical mission information of power transmission lines, mission strategies of drones and historical real-time status data of drones during historical inspections, records of emergencies, manual adjustment strategies and the effects of adjustment strategies; Historical alliance status data is calculated based on historical real-time status data. Historical status data is generated by combining historical alliance status data, historical real-time status data, and emergency event records. Historical action data is generated based on the manual adjustment strategies corresponding to the emergency records. Historical rewards corresponding to the historical action data are calculated based on the effects of the adjustment strategies and the preset reward rules. Based on historical state data, historical action data, and historical rewards, the initial policy adjustment model is trained to obtain the trained policy adjustment model.

[0074] According to one embodiment of the present invention, a storage medium is provided, the storage medium storing at least one executable instruction, which can execute the multi-UAV dynamic task allocation method in any of the above method embodiments.

[0075] Figure 4 The diagram illustrates a structural schematic of a computer device according to an embodiment of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the computer device.

[0076] like Figure 4 As shown, the computer device may include: a processor 402, a communications interface 404, a memory 406, and a communications bus 408.

[0077] The processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408.

[0078] Communication interface 404 is used for communication with other network elements such as clients or other servers.

[0079] The processor 402 is used to execute program 410, specifically to execute the relevant steps in the above-described embodiment of the multi-UAV dynamic task allocation method.

[0080] Specifically, program 410 may include program code that includes computer operation instructions.

[0081] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. A computer device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0082] Memory 406 is used to store program 410. Memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0083] Specifically, program 410 can be used to cause processor 402 to perform the following operations: The inspection task of power transmission lines is divided into multiple task units, and the initial state parameters of each drone are obtained. Based on the initial state parameters of each task unit and each UAV, the alliance is initially divided to obtain multiple initial alliances. The multiple initial alliances are merged and iterated to obtain the optimal alliance. Based on the optimal alliance, the initial mission strategy of each UAV is determined. During the inspection process of each drone according to the initial mission strategy, the real-time status parameters of each drone are obtained, the alliance status parameters are generated based on the real-time status parameters of each drone, the newly added mission priority is obtained, and the occurrence of a sudden event is determined based on the real-time status parameters of each drone and the newly added mission priority. When an emergency occurs, real-time status data is generated based on the emergency, the real-time status parameters of each UAV, and the status parameters of the alliance. The real-time status data is then input into the trained strategy adjustment model to obtain the adjusted task strategy.

[0084] It will be apparent to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. In one embodiment, they can be implemented using device-executable program code, thereby allowing them to be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular hardware and software combination.

[0085] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.

Claims

1. A method for dynamic task allocation among multiple unmanned aerial vehicles (UAVs), characterized in that, include: The inspection task of power transmission lines is divided into multiple task units, and the initial state parameters of each drone are obtained. Based on the initial state parameters of each task unit and each UAV, the alliance is initially divided to obtain multiple initial alliances. The multiple initial alliances are merged and iterated to obtain the optimal alliance. Based on the optimal alliance, the initial task strategy of each UAV is determined. During the inspection process of each UAV according to the initial mission strategy, the real-time status parameters of each UAV are obtained, the alliance status parameters are generated based on the real-time status parameters of each UAV, the newly added mission priority is obtained, and the occurrence of a sudden event is determined based on the real-time status parameters of each UAV and the newly added mission priority. When an emergency occurs, real-time status data is generated based on the emergency, the real-time status parameters of each UAV, and the status parameters of the alliance. The real-time status data is then input into the trained strategy adjustment model to obtain the adjusted task strategy.

2. The multi-UAV dynamic task allocation method as described in claim 1, characterized in that, Based on the initial state parameters of each task unit and each UAV, a preliminary alliance is formed, resulting in multiple initial alliances, including: Based on the task position information of the task unit and the initial position information of the drone, the straight-line distance between each task unit and each drone is calculated, and the idle drone with the closest straight-line distance to the task unit is taken as the single alliance of the drone. Determine whether each single alliance satisfies the boundary constraints. If there is a single alliance that does not satisfy the boundary constraints, reassign the drones corresponding to the single alliance that does not satisfy the boundary constraints and its neighboring drones to the task units until each single alliance satisfies the boundary constraints, thus obtaining multiple initial alliances.

3. The multi-UAV dynamic task allocation method as described in claim 1, characterized in that, The step of merging and iterating the multiple initial alliances to obtain the optimal alliance includes: Calculate the total travel time and total mission time of each drone within the alliance, obtain the initial available time of each drone, and calculate the revenue of a single alliance based on the total travel time, total mission time and initial available time of each drone. Merge any two initial alliances, calculate the total revenue of the merged alliance, and calculate the merged revenue increment based on the total revenue and the revenue of each individual alliance. If the merged revenue increment is greater than zero, the merged alliance is used as the new initial alliance; if the merged revenue increment is less than zero, the original initial alliance is retained. Repeat the above process until there are no two initial alliances whose combined profit increment is greater than zero, thus obtaining the optimal initial alliance.

4. The multi-UAV dynamic task allocation method as described in claim 3, characterized in that, The revenue of a single consortium is calculated using the following method. Where E(S) is the efficiency gain, Q(S) is the quality gain, and B(S) is the equilibrium gain. , , For weights, d j For the drones in the initial alliance j Total movement time within the alliance, c i To complete the mission for drones within the initial alliance i Task time e j This is the initial available time for all drones within the initial alliance; As task priority weight, This is the equipment efficiency coefficient. This represents the average load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance. This represents the maximum load factor of all drones within the initial alliance.

5. The multi-UAV dynamic task allocation method as described in claim 1, characterized in that, The determination of whether a sudden event has occurred based on the real-time status parameters of each UAV and the newly added task priority includes: The fault markers of the UAV are obtained from the real-time status parameters of the UAV. Based on the newly added task priority, task units with a new task priority level greater than a preset level are identified. When the fault marker indicates that a fault has occurred and / or the newly added task priority level is greater than the preset level, an emergency event is determined to have occurred.

6. The multi-UAV dynamic task allocation method as described in claim 1, characterized in that, The alliance status parameters include alliance task completion rate, total task time within the alliance, alliance load difference rate, and proportion of high-priority tasks within the alliance. The generation of alliance status parameters based on the real-time status parameters of each UAV includes: The alliance mission completion rate is calculated based on the number of missions completed by drones within the alliance and the total number of missions included in the alliance. The total time for the alliance mission is the sum of the time taken for each drone to complete its mission and the time taken for it to move. The load rate of a single drone is calculated based on the total mission time of each drone as a percentage of its initial available time. The average load rate of all drones in the alliance is calculated based on the load rates of all single drones. The ratio of the difference between the maximum and minimum load rates of all drones in the alliance to the average load rate of all drones is used as the alliance load difference rate. Calculate the percentage of high-priority tasks within the alliance based on the priority of task units within the alliance.

7. The multi-UAV dynamic task allocation method as described in claim 1, characterized in that, The trained policy adjustment model is obtained using the following methods: Acquire historical mission information of power transmission lines, mission strategies of drones and historical real-time status data of drones during historical inspections, records of emergencies, manual adjustment strategies and the effects of adjustment strategies; Historical alliance status data is calculated based on the historical real-time status data. Historical status data is generated by combining the historical alliance status data, historical real-time status data, and emergency event records. Historical action data is generated based on the manual adjustment strategy corresponding to the emergency record. Historical rewards corresponding to the historical action data are calculated based on the effect of the adjustment strategy and the preset reward rules. Based on the historical state data, historical action data, and historical rewards, the initial policy adjustment model is trained to obtain the trained policy adjustment model.

8. A multi-UAV dynamic task allocation device, characterized in that, include: The initial data acquisition module is used to divide the inspection task of the power transmission line into multiple task units and acquire the initial state parameters of each drone. The initial mission strategy determination module is used to perform preliminary division of the alliance based on the initial state parameters of each mission unit and each UAV, obtain multiple initial alliances, merge and iterate the multiple initial alliances to obtain the optimal alliance, and determine the initial mission strategy of each UAV based on the optimal alliance. The real-time data acquisition module is used to acquire the real-time status parameters of each UAV during the inspection process of each UAV according to the initial task strategy, generate alliance status parameters based on the real-time status parameters of each UAV, acquire the newly added task priority, and determine whether an emergency has occurred based on the real-time status parameters of each UAV and the newly added task priority. The task strategy adjustment module is used to generate real-time status data based on the sudden event, the real-time status parameters of each UAV, and the alliance status parameters when an emergency occurs. The real-time status data is then input into the trained strategy adjustment model to obtain the adjusted task strategy.

9. A storage medium storing at least one executable instruction that causes a processor to perform an operation corresponding to the multi-UAV dynamic task allocation method as described in any one of claims 1-7.

10. A computer device, comprising: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the multi-UAV dynamic task allocation method as described in any one of claims 1-7.

Images