A vehicle-machine collaborative scheduling method for onshore wind farm inspection

Abstract

This invention discloses a vehicle-machine collaborative scheduling method for onshore wind farm inspection, comprising: combining trucks and drones into inspection units as the basic execution units for completing wind farm inspection tasks; acquiring and organizing basic information on wind farm inspection scheduling; establishing a vehicle-machine collaborative scheduling model for onshore wind farm inspection, with the objective function being minimizing the total inspection cost under constraints; using a genetic algorithm based on a neighborhood search strategy to solve the model and obtain the optimal wind farm inspection scheduling scheme; and generating truck driving paths and drone inspection paths based on the optimal scheme to complete the wind farm inspection task. This invention solves the technical problem in existing technologies where truck path planning and drone task allocation fail to coordinate effectively, leading to insufficient rationality in inspection resource allocation, inability to maximize operational efficiency, and consequently increased inspection costs and reduced inspection quality. It has good application prospects and promotional value.

Description

Technical Field

[0001] This invention belongs to the field of intelligent inspection and vehicle-machine collaborative scheduling technology for onshore wind farms, specifically involving a vehicle-machine collaborative scheduling method for onshore wind farm inspection. Background Technology

[0002] As onshore wind farms expand in scale, the efficiency and cost control of their inspection operations have become core challenges for operation and management. Adopting a collaborative inspection model that combines the long-distance transport capabilities of trucks with the flexible inspection advantages of drones is considered an effective way to improve inspection efficiency.

[0003] Currently, most existing technical solutions in this field adopt a "clustering first, planning later" approach. That is, first, the wind turbines are clustered according to their geographical location, and then the drone inspection paths and truck travel paths are optimized in stages. For example, existing technologies have proposed a two-stage wind farm inspection path optimization method, which involves "first step clustering + integer linear programming to solve TSP to optimize UAV path, and second step integer linear programming to solve EQ-GTSP to optimize ground vehicle path" (reference: Baik H, Valenzuela J. An optimization drone routing model for inspecting wind farms [J]. Soft Computing, 2020, 25 (3): 1-16.). Existing technologies involve "first step clustering to divide wind turbines into sub-regions and determine BCS waypoints, and second step integrating 2-OPT local search and simulated annealing acceptance rule-improved GA to solve the UAV vehicle path problem and the BCS traveling salesman problem respectively" (reference: Huang X, Wang G, Lu Y, Jia Z. Study on a Boat-Assisted Drone Inspection Scheme for the Modern Large-Scale Offshore Wind Farm [J]. IEEE Systems Journal, 2023, 17 (3): 4509-4520.).

[0004] However, such technical methods have inherent limitations, resulting in poor performance in practical applications: the fixed clustering partitioning method artificially sets the optimization boundary at the initial stage of the scheme, limiting the possibility of a better match between wind turbines and safety nodes. This rigid structure cannot adapt to complex dynamic constraints such as varying wind turbine inspection times, limited drone endurance, and limited total mission duration, making it difficult to generate refined scheduling schemes.

[0005] Therefore, existing technologies struggle to achieve efficient collaboration between truck and drone resources and overall optimization of inspection costs while meeting various practical constraints. There is an urgent need for an intelligent wind farm inspection scheduling method that can integrate and collaboratively optimize task allocation and path planning, and finely handle various practical constraints, in order to truly improve inspection efficiency, reduce operating costs, and ensure timely operation. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention provides a vehicle-drone collaborative inspection scheduling method for onshore wind farms. This invention addresses the multi-vehicle, multi-drone collaborative inspection scenario in onshore wind farms, constructing a scheduling model that balances drone endurance constraints, truck transportation efficiency, safety node connection requirements, and full inspection coverage requirements. This model more closely reflects the task flow and resource utilization characteristics of "truck-drone" collaborative operations in wind farm inspections. Compared to traditional inspection scheduling research methods that emphasize algorithm complexity, this invention focuses more on the clarity of the model structure and the rationality of the inspection process, emphasizing its feasibility in engineering practice. This method uses the turbine as the basic execution unit for collaborative operations, reflecting the actual inspection process of the wind farm through a concise and effective modeling strategy. It aims to reduce inspection scheduling deployment costs, improve the interpretability of scheduling schemes, and enhance on-site execution efficiency, demonstrating promising application prospects and promotional value. This solves the technical problem in existing inspection scheduling methods where truck route planning and drone task allocation fail to coordinate effectively, leading to insufficient rationality in inspection resource allocation, inability to maximize operational efficiency, and consequently increased inspection costs and reduced inspection quality.

[0007] To achieve the above-mentioned technical objectives, the present invention provides the following technical solution:

[0008] A vehicle-machine collaborative scheduling method for onshore wind farm inspection, specifically including:

[0009] Trucks and drones are combined into inspection units as the basic execution units for completing wind farm inspection tasks; basic information on wind farm inspection scheduling is acquired and organized; the basic information on wind farm inspection scheduling includes the number of units, the number of wind turbines, wind turbine location information, truck travel path information, drone feasible path information, inspection time window, and inspection performance parameters of trucks and drones.

[0010] Based on the obtained basic information on wind farm inspection and scheduling, a vehicle-machine collaborative scheduling model for onshore wind farm inspection is established. Under the constraints, the objective function of the vehicle-machine collaborative scheduling model is to minimize the total inspection cost.

[0011] The optimal wind farm inspection scheduling scheme is obtained by solving the vehicle-machine cooperative scheduling model using a genetic algorithm based on a neighborhood search strategy.

[0012] The optimal wind farm inspection scheduling scheme generates truck driving routes and drone inspection routes to complete the wind farm inspection task.

[0013] Furthermore, the acquisition and organization of basic information for wind farm inspection and scheduling specifically includes:

[0014] The entire wind farm is abstracted into a two-dimensional rectangular coordinate system map, and each wind turbine and safety node in the wind farm is quantitatively represented by a unique coordinate point in this two-dimensional rectangular coordinate system map.

[0015] Obtain the number and coordinates of wind turbines in the wind farm, and denote the set of wind turbines as . ,in Indicates the total number of wind turbines, using Indicates the index number of a single wind turbine;

[0016] Obtain truck travel route information, including the coordinates of safe nodes where trucks stop and the coordinates of inspection bases; and define the set of safe nodes where trucks stop as... ,use An index representing a single secure node. This represents the number of safe nodes, and the set of truck reachable nodes is defined as follows: Index number 0 indicates departure from the inspection base. This indicates a return to the inspection base, defining the set of nodes from which the truck can leave. Define the set of nodes that trucks can drive into. ;

[0017] Obtain the number of inspection units participating in the inspection mission, and denote the set of inspection units as . ,use Indicates the index number of a single inspection unit; each inspection unit It includes a truck and a fixed number of drones; a patrol crew. The drone collection is The index number is ;

[0018] Obtain inspection performance parameters for trucks and drones, including truck data from safety nodes. To the safe node travel time Drones from wind turbines To the fan Flight time Drones targeting wind turbines Inspection time Maximum flight time of drones Fixed cost of each truck unit The waiting cost of a truck per unit of time The operating cost of a truck per unit of time The flight cost of a drone per unit of time Truck traveling at a constant speed The drone's constant flight speed Preparation time for a single takeoff and landing of a drone ;

[0019] Define the set of feasible paths for UAVs based on inspection performance parameters. for:

[0020] ;

[0021] in, This represents an ordered sequence of nodes that indicate the complete inspection path of a drone. This indicates that drones are being assembled for wind turbine inspection. express The modulus length; Indicates the drone's passage A section of the inspection route Flight time;

[0022] Obtain the deadline for wind farm inspection tasks. Start time of wind farm inspection task , That is, the earliest time the truck departs from the inspection base.

[0023] Furthermore, the total inspection cost is calculated based on the inspection route, the inspection performance of the truck and the drone, and includes the fixed cost of the inspection unit, the truck travel cost, the truck waiting cost, and the drone flight cost, expressed by the formula:

[0024] ;

[0025] in, This indicates the total cost of the inspection. Indicates the unit Enable or disable? Indicates the unit Did the trucks come from the node? To the node ; Indicate whether to select a generator set drones Execute patrol path ; For inspection units The truck left the safe node Time; For inspection units The truck reached the safe node The time.

[0026] Furthermore, the constraints of the vehicle-machine collaborative scheduling model for onshore wind farm inspection include:

[0027] The uniqueness constraint for wind turbine inspection ensures that each wind turbine is inspected exactly once.

[0028] The drone inspection path is bound to the inspection object, which restricts the drone to inspecting only the wind turbines on the inspection path it is assigned to.

[0029] The drone inspection path is non-empty, and the constraint is that each drone in the inspection team must perform an inspection operation along at least one path.

[0030] The task start time constraint stipulates that the time when the trucks in each inspection team leave the inspection base shall not be earlier than the start time of the total inspection task.

[0031] When a truck leaves a safety node constraint, if there are drones assigned to inspection paths in the inspection team located at a certain safety node, the truck is constrained to retrieve all drones performing inspection operations before it can leave the safety node.

[0032] The truck can leave the constraint directly when no drone in the inspection team located at a certain safety node is assigned an inspection path;

[0033] The drone takeoff time is restricted, requiring the drone to take off only after the truck has arrived at the safe point and completed takeoff and landing preparations;

[0034] The constraints on drone recovery time include drone takeoff time, drone flight time, fan inspection time, and drone single takeoff and landing preparation time.

[0035] The deadline for inspection tasks is limited to the time when all activated units return to the inspection base, which must not exceed the deadline for the inspection tasks.

[0036] The truck travel continuity constraint ensures that the arrival time of a truck at a safe node is not less than the sum of the departure time at the previous safe node and the truck's travel time.

[0037] Unit departure constraint: The unit must depart from the inspection base to be activated.

[0038] The unit return constraint means that once the constraint is activated, the unit must eventually return to the inspection base.

[0039] The traffic balance constraint for a safe node ensures that the number of truck entry paths is equal to the number of truck exit paths for that safe node.

[0040] MTZ sub-loop constraint elimination;

[0041] The sequential numbering constraint of nodes in the truck's driving path, constraining the sequential numbering of the inspection base as the starting point to be 1;

[0042] The number of inspection paths for drones is limited to a maximum of one inspection path per drone.

[0043] Furthermore, the genetic algorithm based on the neighborhood search strategy is specifically as follows:

[0044] Initialize the input data, which includes the wind turbine set, safety node set, inspection unit set, drone set, coordinates of each wind turbine / safety node, cost coefficient, speed parameter, and inspection / endurance parameter;

[0045] Set the core parameters of the genetic algorithm, including population size, crossover probability, mutation probability, neighborhood search ratio, and maximum number of iterations;

[0046] A two-layer coding mechanism is used to describe the drone task allocation and truck travel route respectively. The upper layer is the drone allocation code, which adopts a tuple list structure. Each tuple in the tuple list corresponds to a complete task description of a drone, which includes in order: take-off and landing safety node number, drone number, and the sequence of wind turbines that the drone needs to inspect. The lower layer is the truck travel route code, which also adopts a tuple list structure. Each tuple in the tuple list corresponds to the travel route of a truck, that is, the safety node numbers that the truck needs to pass through are stored in the access order.

[0047] During decoding, the two-layer encoding is parsed and reconstructed. First, each tuple in the upper-layer encoding is traversed, and the sequence of wind turbines that each UAV needs to inspect is converted from a tuple into a list. Then, it is categorized and integrated according to the take-off and landing safety nodes to construct a UAV allocation dictionary with safety nodes as keys and UAV task mapping table as values. Then, each tuple in the lower-layer encoding is traversed, and the complete path from the inspection base to the safety node and back to the inspection base is completed for each truck to construct a truck driving path dictionary with truck number as key and truck waypoint as value.

[0048] Initialize the population: The population size is initialized based on the genetic algorithm framework, with each individual representing a wind farm inspection scheme. First, a safety node is randomly assigned to each drone. After shuffling the wind turbines, the nodes are traversed and assigned to each drone. Then, an initial path is generated for each truck, containing only the safety nodes used by the drones in the same group. Finally, the drone assignments and truck driving paths are converted into standard individual forms through encoding operations to form the initial population.

[0049] Decode each individual in the population to determine the drone allocation result and truck driving path corresponding to each individual; based on the drone allocation result and truck driving path, cluster the drone inspection tasks in the same unit according to the unit affiliation, and clarify the order of safe nodes that the truck needs to visit in the unit; calculate the total inspection cost of the entire inspection plan, and at the same time perform constraint condition verification, and directly mark the wind farm inspection plan that violates any constraint as an infinite cost plan;

[0050] The total inspection cost is used as the fitness of each individual; the smaller the fitness value, the better the individual.

[0051] Perform hybrid genetic evolution operations: In each generation, selection, crossover, mutation, neighborhood search optimization, and fitness update are performed sequentially to complete the evolutionary search of the population;

[0052] Repeatedly perform fitness calculation and hybrid genetic evolution operation for each individual until the maximum number of iterations is reached, then stop the optimization process; finally output the individual with the smallest fitness value, and after decoding, obtain the globally optimal wind farm inspection scheme.

[0053] Furthermore, the hybrid genetic evolution operation specifically includes:

[0054] Selection operation: A tournament selection strategy is used to select the individuals with the best fitness in each generation of the population to form a new population;

[0055] Crossover operation: Individuals in the new population are paired up, and the binding relationship between the safety node of the parent individual and the drone is randomly inherited according to the drone allocation code. The truck driving path code is executed in an ordered crossover rule according to the truck number to generate offspring individuals.

[0056] Mutation operation: For each offspring individual, randomly fine-tune the list of wind turbines that a certain drone needs to inspect or the sequence of safe nodes that a certain truck needs to visit in sequence, and perform boundary verification on the results, including verification of the drone's endurance, verification of full coverage of wind turbines, and verification of the binding between the drone inspection path and the take-off and landing safe nodes, to ensure that the code is still in the valid space.

[0057] Neighborhood search optimization operation: According to the preset neighborhood search rate, individuals with the highest fitness values ​​are selected and three types of local refinement operations are performed: redistributing the list of wind turbines that the drone needs to inspect within the same unit, replacing the drone take-off and landing safety nodes, and reversing the local driving path of a single truck to generate a better neighborhood solution and replace the original individual, thus completing the individual evolution.

[0058] The fitness update operation re-decodes, verifies boundaries, and recalculates the fitness of evolved individuals, updating the current best individual and the global best solution in the population.

[0059] Furthermore, the present invention also proposes an electronic device comprising a memory and a processor, wherein:

[0060] Memory is used to store computer programs that can run on a processor;

[0061] The processor is used to execute, as described above, a vehicle-machine collaborative scheduling method for onshore wind farm inspection when running the computer program.

[0062] A computer-readable storage medium is also proposed, which stores computer instructions for causing a processor to execute a vehicle-machine collaborative scheduling method for onshore wind farm inspection as described above.

[0063] Based on the above technical solution, the present invention has at least the following beneficial effects:

[0064] 1. This invention addresses the vehicle-turbine collaborative inspection scenario of onshore wind farms and constructs a scheduling model that takes into account the constraints of drone endurance, truck transportation efficiency, safety node connection requirements, and wind turbine full coverage requirements. It can more accurately reflect the task flow and resource load characteristics in the actual inspection of wind farms, and effectively improve the executability of the inspection scheduling scheme and the system response speed.

[0065] 2. To enhance the scheduling model's ability to match inspection operation rhythm, this invention, based on the integrated modeling of truck route planning and UAV task allocation, further introduces an inspection timing coordination model based on multi-constraint coupling. This model comprehensively considers factors such as UAV take-off and landing preparation time, wind turbine inspection duration, and truck waiting connection time to construct more precise inspection operation time window constraints, thereby dynamically correcting the feasible scheduling space and ensuring the coordination and accuracy between truck and UAV operation links. Simultaneously, through the orderly connection of unit-level coordination strategies and safety node dwell time sequences, while meeting the constraints of UAV endurance, truck task deadlines, and inspection resources, the rationality of inspection task organization and operational efficiency are improved, further enhancing the executability of the scheduling scheme and its fit with actual inspection operations.

[0066] 3. The design concept of the vehicle-machine collaborative inspection and scheduling scenario proposed in this invention is simple and logically clear, which facilitates the connection with the actual inspection process of wind farms, reduces the deployment cost of the scheduling system, improves the interpretability of the solution, and can be quickly applied in the actual wind farm inspection environment. It has good engineering practice value and promotion potential. Attached Figure Description

[0067] Figure 1 This is a flowchart of a vehicle-machine collaborative scheduling method for onshore wind farm inspection proposed in this invention;

[0068] Figure 2This is a schematic diagram of an onshore wind farm inspection scenario for which the method proposed in this invention is intended.

[0069] Figure 3 This is a flowchart of the genetic algorithm based on the neighborhood search strategy used in the method proposed in this invention to solve the optimal scheduling scheme. Detailed Implementation

[0070] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the following description is provided in conjunction with the accompanying drawings. Figures 1-3 The present invention will be further described in detail with reference to the embodiments. This will allow for a full understanding of how the present application uses technical means to solve technical problems and achieve technical effects, and to facilitate its implementation.

[0071] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0072] Please refer to Figure 1 This embodiment proposes a vehicle-drone collaborative inspection scheduling method for onshore wind farms. By rationally allocating wind turbine inspection tasks between trucks and drones, and combining a linkage processing mechanism of unit-level collaborative strategy and path timing optimization, it not only effectively improves the utilization efficiency of inspection resources and the overall execution efficiency of tasks, but also provides a practical and rapidly deployable scheduling solution for onshore wind farm inspection operations. The method includes the following steps:

[0073] S1. Combine trucks and drones into inspection units as the basic execution unit to complete wind farm inspection tasks; acquire and organize basic information on wind farm inspection scheduling; the basic information on wind farm inspection scheduling includes the number of units, the number of wind turbines, wind turbine location information, truck driving path information, drone feasible path information, inspection time window, and inspection performance parameters of trucks and drones.

[0074] In a preferred embodiment, the acquisition and organization of basic information on wind farm inspection and scheduling specifically includes:

[0075] The entire wind farm is abstracted into a two-dimensional rectangular coordinate system map, and each wind turbine and safety node in the wind farm is quantitatively represented by a unique coordinate point in this two-dimensional rectangular coordinate system map.

[0076] Obtain the number and coordinates of wind turbines in the wind farm, and denote the set of wind turbines as . ,in Indicates the total number of wind turbines, using Indicates the index number of a single wind turbine;

[0077] Obtain truck travel route information, including the coordinates of safe nodes where trucks stop and the coordinates of inspection bases; and define the set of safe nodes where trucks stop as... ,use An index representing a single secure node. This represents the number of safe nodes, and the set of truck reachable nodes is defined as follows: Index number 0 indicates departure from the inspection base. This indicates a return to the inspection base, defining the set of nodes from which the truck can leave. Define the set of nodes that trucks can drive into. ;

[0078] Obtain the number of inspection units participating in the inspection mission, and denote the set of inspection units as . ,use Indicates the index number of a single inspection unit; each inspection unit It includes a truck and a fixed number of drones; a patrol crew. The drone collection is The index number is ;

[0079] Obtain inspection performance parameters for trucks and drones, including truck data from safety nodes. To the safe node travel time (Manhattan distance calculation based on roads), drones from wind turbines To the fan Flight time (Based on Euclidean distance calculations), drones targeting wind turbines Inspection time Maximum flight time of drones Fixed cost of each truck unit The waiting cost of a truck per unit of time The operating cost of a truck per unit of time The flight cost of a drone per unit of time Truck traveling at a constant speed The drone's constant flight speed Preparation time for a single takeoff and landing of a drone (The drone's takeoff and landing are timed once each);

[0080] Define the set of feasible paths for UAVs based on inspection performance parameters. for:

[0081] ;

[0082] in, This represents an ordered sequence of nodes representing the complete inspection path of a drone, i.e., the path from which the drone starts at a safe node. Take off and inspect the wind turbines in sequence. Finally, return to the safe node. landing; This indicates that drones are being assembled for wind turbine inspection. express The modulus length; that is The meaning is that the unit drones in At the secure node Takeoff and landing, sequential inspection and assembly. The fan in the middle; Indicates the drone's passage A section of the inspection route The flight time can be further broken down as follows:

[0083] ;

[0084] for , Indicates that the drone departed from the secure node arrive China's first wind turbine Flight time, Indicates from the wind turbine To the next wind turbine Flight time, Indicates from The last wind turbine Return to safe node Flight time.

[0085] Obtain the deadline for wind farm inspection tasks. Start time of wind farm inspection task , That is, the earliest time the truck departs from the inspection base.

[0086] S2. Based on the obtained basic information on wind farm inspection and scheduling, the wind farm inspection scenario can be as follows: Figure 2 As shown, a vehicle-machine collaborative scheduling model for onshore wind farm inspection is established. Under the constraints, the objective function of the vehicle-machine collaborative scheduling model is to minimize the total inspection cost.

[0087] In a preferred embodiment, the total inspection cost is calculated based on the inspection route, the inspection performance of the truck and the drone, and includes the fixed cost of the inspection unit, the truck travel cost, the truck waiting cost, and the drone flight cost, expressed by the following formula:

[0088] ;

[0089] in, This indicates the total cost of the inspection. Indicates the unit Enable or disable? Indicates the unit Did the trucks come from the node? To the node ; Indicate whether to select a generator set drones Execute inspection route ; For inspection units The truck left the safe node Time; For inspection units The truck reached the safe node The time.

[0090] Regarding the constraints of the vehicle-machine collaborative scheduling model for onshore wind farm inspection, this embodiment first sets the following assumptions:

[0091] A fixed number of drones are carried on a truck. The speed of the drones is greater than that of the truck, and each truck and each drone is homogeneous. Drones are released and recovered at the same safe node where the truck is parked. The truck must wait for all drones to be recovered before proceeding to the next node. There is no limit to the number of drones that can be taken off or recovered at the same time. It is assumed that the space conditions allow and no collisions will occur. The drone take-off and landing time plus the battery swapping time are all included in the drone take-off and landing preparation time parameter. Each truck and drone in each unit is fixed and cannot be recovered across different trucks.

[0092] Based on the above assumptions, in this embodiment, the constraints of the vehicle-machine collaborative scheduling model for onshore wind farm inspection include:

[0093] The uniqueness constraint for wind turbine inspection ensures that each wind turbine is inspected exactly once, expressed by the formula:

[0094] ;

[0095] The drone inspection path is bound to the inspection object, which restricts the drone to inspecting only the wind turbines on its assigned inspection path. The formula is as follows:

[0096] ;

[0097] The drone inspection path is non-empty, meaning that each drone in the activated inspection team must perform inspection operations along at least one path. The formula is as follows:

[0098] ;

[0099] The mission time constraint sets the time for each inspection crew's trucks to leave the inspection base. Not earlier than the start time of the overall inspection task, expressed by the formula:

[0100] ;

[0101] When a truck leaves a safety node, if there are drones assigned to inspection paths in the inspection team located at a certain safety node, the truck is constrained to only leave the safety node after recovering all drones performing inspection operations. The formula is as follows:

[0102] ;

[0103] The truck can leave the constraint directly when no drone in the inspection team located at a certain safety node has been assigned an inspection path. The formula is as follows:

[0104] ;

[0105] The drone takeoff time constraint requires that the drone can only take off after the truck has arrived at the safe node and completed takeoff and landing preparations. The formula is as follows:

[0106] ;

[0107] The drone recovery time constraint includes the drone takeoff time, drone flight time, fan inspection time, and drone single takeoff and landing preparation time, expressed by the formula:

[0108] It should be noted that "here" This indicates the premise for the drone recovery time calculation formula to hold true;

[0109] Inspection task deadline constraints, which limit the time required for all activated units to return to the inspection base. The deadline for the inspection task shall not be exceeded, as expressed in the formula:

[0110] ;

[0111] The truck travel continuity constraint ensures that the arrival time of a truck at a safe node is not less than the sum of its departure time at the previous safe node and its travel time. The formula is as follows:

[0112] ;

[0113] Unit departure constraint: The constraint requires that the unit must depart from the inspection base to be activated. The formula is expressed as follows:

[0114] ;

[0115] Unit return constraint: With the constraint enabled, the unit must ultimately return to the inspection base. The formula is as follows:

[0116] ;

[0117] The traffic balance constraint for a safe node ensures that the number of truck entry paths equals the number of truck exit paths. The formula is as follows:

[0118] ;

[0119] Here For three different security nodes;

[0120] The MTZ sub-loop constraint elimination formula is expressed as follows:

[0121] ;

[0122] The sequential numbering constraint of nodes in the truck's driving path, constraining the sequential numbering of the inspection bases at the starting point. The value is 1, and the formula is expressed as:

[0123] ;

[0124] The constraint on the number of inspection paths for drones limits each drone to selecting at most one inspection path, as expressed by the formula:

[0125] .

[0126] Furthermore, this embodiment also provides the following value range constraints for each parameter involved in the above formula:

[0127] ;

[0128] ;

[0129] ;

[0130] ;

[0131] ;

[0132] ;

[0133] ;

[0134] ;

[0135] ;

[0136] in, Indicates the unit drones Are the fans inspected? ; Indicate whether to select a generator set drones Execution path ; For the unit The truck reached the safe node Time; For the unit The truck left the safe node Time; For the unit drones From security nodes Departure time; For the unit drones Return to safe node Time; Indicates the unit Enable or disable? Indicates the unit Did the truck leave the safe node? To the safe node ; For the unit Truck driving path safety nodes The sequential numbering.

[0137] S3. Use a genetic algorithm based on a neighborhood search strategy to solve the vehicle-machine cooperative scheduling model and obtain the optimal wind farm inspection scheduling scheme.

[0138] As a preferred embodiment, in this embodiment, such as Figure 3 As shown, the genetic algorithm based on the neighborhood search strategy is specifically as follows:

[0139] Initialize the input data, which includes the wind turbine set, safety node set, inspection unit set, drone set, coordinates of each wind turbine / safety node, cost coefficient, speed parameter, and inspection / endurance parameter (cost coefficient, speed parameter, and inspection / endurance parameter are used to calculate various inspection costs);

[0140] Set the core parameters of the genetic algorithm, including population size, crossover probability, mutation probability, neighborhood search ratio, and maximum number of iterations;

[0141] A two-layer coding mechanism is used to describe the drone task allocation and truck travel route respectively. The upper layer is the drone allocation code, which adopts a tuple list structure. Each tuple in the tuple list corresponds to a complete task description of a drone, which includes in order: take-off and landing safety node number, drone number, and the sequence of wind turbines that the drone needs to inspect. The lower layer is the truck travel route code, which also adopts a tuple list structure. Each tuple in the tuple list corresponds to the travel route of a truck, that is, the safety node numbers that the truck needs to pass through are stored in the access order.

[0142] During decoding, the two-layer encoding is parsed and reconstructed. First, each tuple in the upper layer encoding is traversed, and the sequence of wind turbines that each drone needs to inspect is converted from a tuple into a list. Then, it is categorized and integrated according to the take-off and landing safety nodes to construct a drone allocation dictionary with safety nodes as keys and drone task mapping table (i.e., the wind turbines that each drone needs to inspect) as values. Then, each tuple in the lower layer encoding is traversed, and the complete path for each truck from the inspection base to passing through the safety nodes and returning to the inspection base is completed. That is, a starting point (starting from the inspection base) is added to the beginning of the tuple sequence, and an ending point (returning to the inspection base) is added to the end of the tuple sequence to construct a truck driving path dictionary with truck number as key and truck waypoints (including starting point, ending point, and passing through safety nodes) as values.

[0143] The two-layer encoding fully expresses the task allocation and path planning information in the form of a compact list of tuples; then the decoding process transforms the encoded information into a directly executable task allocation scheme and path navigation scheme.

[0144] Initialize the population: The population size is initialized based on the genetic algorithm framework, with each individual representing a wind farm inspection scheme. First, a safety node is randomly assigned to each drone. After shuffling the wind turbines, the nodes are traversed and assigned to each drone. Then, an initial path is generated for each truck, containing only the safety nodes used by the drones in the same group. Finally, the drone assignments and truck driving paths are converted into standard individual forms through encoding operations to form the initial population.

[0145] Decode each individual in the population to determine the drone allocation result and truck driving path corresponding to each individual; based on the drone allocation result and truck driving path, cluster the drone inspection tasks in the same unit according to the unit affiliation, and clarify the order of safe nodes that the truck needs to visit in the unit; calculate the total inspection cost of the entire inspection plan, and at the same time perform constraint condition verification, and directly mark the wind farm inspection plan that violates any constraint as an infinite cost plan;

[0146] The total inspection cost is used as the fitness of each individual; the smaller the fitness value, the better the individual.

[0147] Perform hybrid genetic evolution operations: In each generation, selection, crossover, mutation, neighborhood search optimization, and fitness update are performed sequentially to complete the evolutionary search of the population;

[0148] The specific process of hybrid genetic evolution operations includes:

[0149] Selection operation: A tournament selection strategy is used to select the individuals with the best fitness in each generation of the population to form a new population;

[0150] Crossover operation: Individuals in the new population are paired up, and the binding relationship between the safety node of the parent individual and the drone is randomly inherited according to the drone allocation code. The truck driving path code is executed in an orderly crossover rule according to the truck number (i.e. the crew number) to generate offspring individuals.

[0151] Mutation operation: For each offspring individual, randomly fine-tune the list of wind turbines that a certain drone needs to inspect or the sequence of safe nodes that a certain truck needs to visit in sequence. Perform boundary verification on the results, including verification of the drone's endurance, verification of full coverage of wind turbines, and verification of the binding between the drone's inspection path and the take-off and landing safe nodes, to ensure that the code is still in the valid space.

[0152] Neighborhood search optimization operation: According to the preset neighborhood search rate, individuals with the highest fitness values ​​are selected and three types of local refinement operations are performed: redistributing the list of wind turbines that the drone needs to inspect within the same unit, replacing the drone take-off and landing safety nodes, and reversing the local driving path of a single truck to generate a better neighborhood solution and replace the original individual, thus completing the individual evolution.

[0153] The fitness update operation re-decodes, verifies boundaries, and calculates fitness for the evolved individuals, updating the current best individual and the global best solution in the population.

[0154] Repeatedly perform fitness calculation and hybrid genetic evolution operation for each individual until the maximum number of iterations is reached, then stop the optimization process; finally output the individual with the smallest fitness value, and after decoding, obtain the globally optimal wind farm inspection scheme.

[0155] S4. Generate truck driving routes and drone inspection routes based on the optimal wind farm inspection scheduling scheme to complete the wind farm inspection task.

[0156] Furthermore, the present invention also proposes an electronic device comprising a memory and a processor, wherein:

[0157] Memory is used to store computer programs that can run on a processor;

[0158] The processor is used to execute, as described above, a vehicle-machine collaborative scheduling method for onshore wind farm inspection when running the computer program.

[0159] A computer-readable storage medium is also proposed, which stores computer instructions for causing a processor to execute a vehicle-machine collaborative scheduling method for onshore wind farm inspection as described above.

[0160] This concludes the description of the entire process of the vehicle-turbine collaborative scheduling method for onshore wind farm inspection proposed in this invention. The following embodiment will provide a specific example of multi-unit inspection in an onshore wind farm, verifying the vehicle-turbine collaborative inspection scheduling in conjunction with the method proposed in this invention.

[0161] A specific example is set up as follows: An onshore wind farm needs to complete the inspection of 20 wind turbines. Three inspection teams (numbered 0, 1, and 2) are configured. Each team corresponds to one truck, and each team is equipped with two inspection drones (the six drones are numbered 0-5, and the mapping relationship between the teams and drones is as follows: Team 0 corresponds to drones 0 and 1, Team 1 corresponds to drones 2 and 3, and Team 2 corresponds to drones 4 and 5).

[0162] Inspection operations are represented by node 0, indicating the start from the inspection base, and node 29, indicating the return to the inspection base. There are 8 safe nodes for drone take-off and landing (i.e., safe nodes where trucks can park) in the wind farm. The specific coordinates of all wind turbines to be inspected, safe nodes, and inspection bases are shown in Table 1 below.

[0163] Table 1. Coordinates of each node

[0164] Node type Node number coordinate Inspection Base (Departure) 0 (0, 0) Inspection Base (Return) 29 (0, 0) Fan 1 (4, 6) Fan 2 (10, 8) Fan 3 (8, 2) Fan 4 (14, 4) Fan 5 (18, 10) Fan 6 (12, 14) Fan 7 (2, 12) Fan 8 (6, 16) Fan 9 (16, 2) Fan 10 (8, 10) Fan 11 (36, 9) Fan 12 (45, 12) Fan 13 (42, 3) Fan 14 (51, 6) Fan 15 (57, 15) Fan 16 (48, 21) Fan 17 (33, 18) Fan 18 (39, 24) Fan 19 (54, 3) Fan 20 (42, 15) Security Node 21 (2, 4) Security Node 22 (6, 6) Security Node 23 (33, 6) Security Node 24 (39, 9) Security Node 25 (10, 24)

[0165] The operating cost and speed parameters are set as follows: fixed cost of unit usage: 1000 yuan / vehicle; unit cost of truck travel: 100 yuan / hour (travel speed 40 km / h, mileage calculated based on Manhattan distance); unit cost of drone flight: 200 yuan / hour (flight speed 80 km / h, mileage calculated based on Euclidean distance); unit cost of truck waiting: 100 yuan / hour. The basic inspection time for each wind turbine varies (see Table 2 below for specific times); the preparation time for a single drone takeoff and landing is 1 / 12 hour, with a maximum endurance of 3 hours; the total mission deadline for each truck from departure to return to base is 24 hours.

[0166] Table 2 Time Consumption for Inspection of Each Fan Foundation

[0167] Fan number Inspection time (hours) 1 0.80 2 0.75 3 0.70 4 0.78 5 0.85 6 0.72 7 0.90 8 0.68 9 0.78 10 0.72 11 0.80 12 0.75 13 0.70 14 0.78 15 0.85 16 0.72 17 0.90 18 0.68 19 0.78 20 0.72

[0168] Based on the above settings, a scheduling scheme is designed according to the method proposed in this invention, planning the truck's driving route and the UAV's take-off and landing sequence, minimizing the total inspection cost while satisfying all constraints. The final output results are as follows:

[0169] [Total Cost Summary]

[0170] Optimal total cost: 8399.64 yuan

[0171] Fixed cost of trucks (2 vehicles): 2000.00 yuan

[0172] Truck travel and waiting costs: 1652.46 yuan

[0173] Total cost of drone inspection: 4747.18 yuan

[0174] Total Completion Time

[0175] Completion time for each truck in use:

[0176] Truck 0: 11.091 hours

[0177] Truck 1: 5.434 hours

[0178] Total project completion time: 11.091 hours

[0179] Is it less than the deadline (24 hours): Yes

[0180] [Truck Route and Drone Mission Details]

[0181] Truck 0:

[0182] Path: 0 → 23 → 24 → 22 → 29

[0183] Node 23:

[0184] Drone 0 → Inspecting wind turbines [9,15] (total of 2 units) | Time taken: 2.873 hours (≤3 hours of flight time)

[0185] Drone 1 → Inspecting wind turbines [17,13] (2 units total) | Time taken: 2.254 hours (≤3 hours flight time)

[0186] Node 24:

[0187] Drone 0 → Inspecting wind turbines [12,14,20] (total of 3 units) | Time taken: 2.850h (≤3h flight time)

[0188] Drone 1 → Inspecting wind turbines [2,11] (total of 2 units) | Time taken: 2.442 hours (≤3 hours of flight time)

[0189] Node 22:

[0190] Drone 0 → Inspecting wind turbines [8,1,7] (total 3 units) | Time taken: 2.968h (≤3h flight time)

[0191] Drone 1 → Inspecting wind turbines [16,5] (2 units total) | Time taken: 2.852 hours (≤3 hours of flight time)

[0192] Truck 1:

[0193] Path: 0 → 21 → 22 → 29

[0194] Node 21:

[0195] Drone 3 → Inspecting wind turbines

[18] (1 unit in total) | Time taken: 1.898h (≤3h flight time)

[0196] Node 22:

[0197] Drone 2 → Inspecting wind turbines [10,4,3] (total of 3 units) | Time taken: 2.664 hours (≤3 hours of flight time)

[0198] Drone 3 → Inspecting wind turbines [19,6] (2 units total) | Time taken: 2.936 hours (≤3 hours flight time)

[0199] Truck 2:

[0200] Status: Not enabled

[0201] Therefore, the scheduling scheme obtained by this method can minimize the total inspection cost and complete the inspection task within the specified inspection time window.

[0202] In summary, this invention addresses the vehicle-turbine collaborative inspection scenario in onshore wind farms by proposing an integrated vehicle-turbine collaborative scheduling method. First, it collects wind turbine coordinates, truck parking points, inspection time windows, and drone performance data. Based on this, a mathematical model is established with the goal of minimizing inspection costs. A genetic algorithm based on a neighborhood search strategy is designed for double-layer encoding and solving, rapidly generating truck paths and drone takeoff sequences. This method significantly reduces the total inspection cost and improves wind farm operation and maintenance efficiency. The model has a simple structure and clear constraints, can be directly integrated with existing roads and flight control systems, has low deployment costs, strong interpretability, and possesses good engineering practice value and promotion potential.

[0203] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0204] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for vehicle-machine cooperative scheduling for land-based wind farm inspection, characterized in that, Specifically, the following steps are included: Trucks and drones are combined into inspection units to serve as the basic execution units for wind farm inspection tasks; basic information on wind farm inspection scheduling is acquired and organized; there are safe take-off and landing nodes for drones within the wind farm, which are also safe nodes where trucks can dock; the basic information on wind farm inspection scheduling includes the number of units, the number of wind turbines, wind turbine location information, truck travel path information, feasible path information for drones, inspection time windows, and inspection performance parameters of trucks and drones; Based on the obtained basic information on wind farm inspection and scheduling, a vehicle-machine collaborative scheduling model for onshore wind farm inspection is established. Under the constraints, the objective function of the vehicle-machine collaborative scheduling model is to minimize the total inspection cost. A genetic algorithm based on a neighborhood search strategy is used to solve the vehicle-machine cooperative scheduling model to obtain the optimal wind farm inspection scheduling scheme; the genetic algorithm based on the neighborhood search strategy specifically includes: Initialize the input data, which includes the wind turbine set, safety node set, inspection unit set, drone set, coordinates of each wind turbine, coordinates of the safety node, cost coefficient, speed parameter, inspection parameter, and endurance parameter; Set the core parameters of the genetic algorithm, including population size, crossover probability, mutation probability, neighborhood search ratio, and maximum number of iterations; A two-layer coding mechanism is used to describe the drone task allocation and truck travel route respectively. The upper layer is the drone allocation code, which adopts a tuple list structure. Each tuple in the tuple list corresponds to a complete task description of a drone, which includes in order: take-off and landing safety node number, drone number, and the sequence of wind turbines that the drone needs to inspect. The lower layer is the truck travel route code, which also adopts a tuple list structure. Each tuple in the tuple list corresponds to the travel route of a truck, that is, the safety node numbers that the truck needs to pass through are stored in the access order. During decoding, the two-layer encoding is parsed and reconstructed. First, each tuple in the upper-layer encoding is traversed, and the sequence of wind turbines that each UAV needs to inspect is converted from a tuple into a list. Then, it is categorized and integrated according to the take-off and landing safety nodes to construct a UAV allocation dictionary with safety nodes as keys and UAV task mapping table as values. Then, each tuple in the lower-layer encoding is traversed, and the complete path from the inspection base to the safety node and back to the inspection base is completed for each truck to construct a truck driving path dictionary with truck number as key and truck waypoint as value. The optimal wind farm inspection scheduling scheme generates truck driving routes and drone inspection routes to complete the wind farm inspection task. 2.The method of claim 1, wherein, The acquisition and organization of basic information for wind farm inspection and scheduling specifically includes: The entire wind farm is abstracted into a two-dimensional rectangular coordinate system map, and each wind turbine and safety node in the wind farm is quantitatively represented by a unique coordinate point in this two-dimensional rectangular coordinate system map. Obtain the number and coordinates of wind turbines in the wind farm, and denote the wind turbine set as wherein represents the total number of wind turbines, and uses to represent the index number of a single wind turbine; Obtain truck driving path information, including the coordinates of the safety nodes where the truck stops and the coordinates of the inspection base; and define the set of safety nodes where the truck stops as , use to represent the index of a single safety node, to represent the number of safety nodes, and define the set of reachable nodes of the truck as , where index number 0 represents departure from the inspection base, index number represents return to the inspection base, define the set of drivable nodes of the truck as , and define the set of drivable nodes of the truck as ; Obtain the number of inspection units participating in the inspection mission, and denote the set of inspection units as . ,use Indicates the index number of a single inspection unit; each inspection unit It includes a truck and a fixed number of drones; a patrol crew. The drone collection is The index number is ; Obtain inspection performance parameters for trucks and drones, including truck data from safety nodes. To the safe node travel time Drones from wind turbines To the fan Flight time Drones targeting wind turbines Inspection time Maximum flight time of drones Fixed cost of each truck unit The waiting cost of a truck per unit of time The operating cost of a truck per unit of time The flight cost of a drone per unit of time Truck traveling at a constant speed The drone's constant flight speed Preparation time for a single takeoff and landing of a drone ; Define the set of feasible paths for UAVs based on inspection performance parameters. for: ； in, This represents an ordered sequence of nodes that indicate the complete inspection path of a drone. This indicates that drones are being assembled for wind turbine inspection. express The modulus length; Indicates the drone's passage A section of the inspection route Flight time; Obtain the deadline for wind farm inspection tasks. Start time of wind farm inspection task , That is, the earliest time the truck departs from the inspection base.

3. The vehicle-machine collaborative scheduling method for onshore wind farm inspection according to claim 2, characterized in that, The total inspection cost is calculated based on the inspection route, the inspection performance of trucks and drones, and includes the fixed cost of the inspection team, the truck travel cost, the truck waiting cost, and the drone flight cost. The formula is as follows: ； in, This indicates the total cost of the inspection. Indicates the unit Enable or disable? Indicates the unit Did the trucks come from the node? To the node ; Indicate whether to select a generator set drones Execute inspection route ; For inspection units The truck left the safe node Time; For inspection units The truck reached the safe node The time.

4. The vehicle-machine collaborative scheduling method for onshore wind farm inspection according to claim 2, characterized in that, The constraints of the vehicle-machine collaborative scheduling model for onshore wind farm inspection include: The uniqueness constraint for wind turbine inspection ensures that each wind turbine is inspected exactly once. The drone inspection path is bound to the inspection object, which restricts the drone to inspecting only the wind turbines on the inspection path it is assigned to. The drone inspection path is non-empty, and the constraint is that each drone in the inspection team must perform an inspection operation along at least one path. The task start time constraint stipulates that the time when the trucks in each inspection team leave the inspection base shall not be earlier than the start time of the total inspection task. When a truck leaves a safety node constraint, if there are drones assigned to inspection paths in the inspection team located at a certain safety node, the truck is constrained to retrieve all drones performing inspection operations before it can leave the safety node. The truck can leave the constraint directly when no drone in the inspection team located at a certain safety node is assigned an inspection path; The drone takeoff time is restricted, requiring the drone to take off only after the truck has arrived at the safe point and completed takeoff and landing preparations; The constraints on drone recovery time include drone takeoff time, drone flight time, fan inspection time, and drone single takeoff and landing preparation time. The deadline for inspection tasks is limited to the time when all activated units return to the inspection base, which must not exceed the deadline for the inspection tasks. The truck travel continuity constraint ensures that the arrival time of a truck at a safe node is not less than the sum of the departure time at the previous safe node and the truck's travel time. Unit departure constraint: The unit must depart from the inspection base to be activated. The unit return constraint means that once the constraint is activated, the unit must eventually return to the inspection base. The traffic balance constraint for a safe node ensures that the number of truck entry paths is equal to the number of truck exit paths for that safe node. MTZ sub-loop constraint elimination; The sequential numbering constraint of nodes in the truck's driving path, constraining the sequential numbering of the inspection base as the starting point to be 1; The number of inspection paths for drones is limited to a maximum of one inspection path per drone.

5. The vehicle-machine collaborative scheduling method for onshore wind farm inspection according to claim 1, characterized in that, The genetic algorithm based on the neighborhood search strategy also includes: Initialize the population: The population size is initialized based on the genetic algorithm framework, with each individual representing a wind farm inspection scheme. First, a safety node is randomly assigned to each drone. After shuffling the wind turbines, the nodes are traversed and assigned to each drone. Then, an initial path is generated for each truck, containing only the safety nodes used by the drones in the same group. Finally, the drone assignments and truck driving paths are converted into standard individual forms through encoding operations to form the initial population. Decode each individual in the population to determine the drone allocation result and truck driving path corresponding to each individual; based on the drone allocation result and truck driving path, cluster the drone inspection tasks in the same unit according to the unit affiliation, and clarify the order of safe nodes that the truck needs to visit in the unit; calculate the total inspection cost of the entire inspection plan, and at the same time perform constraint condition verification, and directly mark the wind farm inspection plan that violates any constraint as an infinite cost plan; The total inspection cost is used as the fitness of each individual; the smaller the fitness value, the better the individual. Perform hybrid genetic evolution operations: In each generation, selection, crossover, mutation, neighborhood search optimization, and fitness update are performed sequentially to complete the evolutionary search of the population; Repeatedly perform fitness calculation and hybrid genetic evolution operation for each individual until the maximum number of iterations is reached, then stop the optimization process; finally output the individual with the smallest fitness value, and after decoding, obtain the globally optimal wind farm inspection scheme.

6. The vehicle-machine collaborative scheduling method for onshore wind farm inspection according to claim 5, characterized in that, The hybrid genetic evolution operation specifically involves: Selection operation: A tournament selection strategy is used to select the individuals with the best fitness in each generation of the population to form a new population; Crossover operation: Individuals in the new population are paired up, and the binding relationship between the safety node of the parent individual and the drone is randomly inherited according to the drone allocation code. The truck driving path code is executed in an ordered crossover rule according to the truck number to generate offspring individuals. Mutation operation: For each offspring individual, randomly fine-tune the list of wind turbines that a certain drone needs to inspect or the sequence of safe nodes that a certain truck needs to visit in sequence, and perform boundary verification on the results, including verification of the drone's endurance, verification of full coverage of wind turbines, and verification of the binding between the drone inspection path and the take-off and landing safe nodes, to ensure that the code is still in the valid space. Neighborhood search optimization operation: According to the preset neighborhood search rate, individuals with the highest fitness values ​​are selected and three types of local refinement operations are performed: redistributing the list of wind turbines that the drone needs to inspect within the same unit, replacing the drone take-off and landing safety nodes, and reversing the local driving path of a single truck to generate a better neighborhood solution and replace the original individual, thus completing the individual evolution. The fitness update operation re-decodes, verifies boundaries, and recalculates the fitness of evolved individuals, updating the current best individual and the global best solution in the population.

7. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein: Memory is used to store computer programs that can run on a processor; A processor is configured to, when running the computer program, execute a vehicle-machine collaborative scheduling method for onshore wind farm inspection as described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause the processor to execute a vehicle-machine collaborative scheduling method for onshore wind farm inspection as described in any one of claims 1-6.

Images