A low-altitude drone group scheduling method and system based on a transit search algorithm and a single drone power-changing mother ship escorting

Abstract

This invention proposes a scheduling method and system for a low-altitude UAV swarm with a single UAV battery-swapping carrier escort, based on a transit search algorithm, relating to the field of UAV swarm scheduling and energy management technology. This invention achieves breakthrough improvements in endurance, environmental adaptability, and collaborative efficiency through spatiotemporal coupling constraint decoupling and a dynamic battery-swapping decision mechanism. The method includes: UAV swarm task input and information acquisition; determining the scheduling decision variables for the UAV swarm with a single UAV battery-swapping carrier escort; determining the scheduling objective function and related constraints for the UAV swarm with a single UAV battery-swapping carrier escort; establishing a low-altitude UAV swarm scheduling model with a single UAV battery-swapping carrier escort based on the determined objective function and its constraints; and solving the scheduling model for the UAV swarm with a single UAV battery-swapping carrier escort using a machine learning-based improved transit search algorithm. This invention is applicable to UAV swarm scheduling and energy management.

Description

A scheduling method and system for low-altitude UAV swarms including single UAV battery swapping carrier escort based on transit search algorithm Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) swarm scheduling and energy management technology, and in particular to a scheduling method for a low-altitude UAV swarm including a single UAV battery swapping carrier accompanied by a carrier, based on a transit search algorithm. Background Technology

[0002] Against the strategic backdrop of the rapid rise of the low-altitude economy, drones, as the main operators of low-altitude airspace, have expanded their application scenarios from traditional geographic surveying to urban air traffic, logistics delivery, disaster emergency rescue, and other fields. However, existing drone swarm scheduling technologies face several bottlenecks: First, drones are limited by their endurance in long-duration and long-distance missions. Existing static task allocation algorithms, lacking a dynamic charging and battery swapping scheduling mechanism, result in a high interruption rate for long-endurance missions. Second, they lack adaptability to complex dynamic environments. Traditional heuristic algorithms suffer from reduced real-time path planning under multi-dimensional interference such as sudden airspace control and weather disturbances, making it difficult to meet high-precision requirements. Third, current technology fails to fully utilize the battery swapping resources of the accompanying carrier, leading to resource waste and poor scheduling efficiency. Furthermore, the collaborative operation mechanism between drones and the accompanying carrier is also lacking, affecting overall mission execution efficiency. It is worth noting that existing technical documents do not yet present a complete scheduling solution for using escort carriers for battery swapping of UAV swarms, and in particular, they do not propose a scheduling framework for integrated collaborative path optimization and dynamic matching between escort carriers and UAV swarms. These issues need to be addressed through improved algorithms and optimized resource management to enhance the scheduling efficiency and mission execution capabilities of UAV swarms. This scheduling optimization not only improves the operational efficiency of UAVs and fills the technological gap in the field of dynamic charging and swapping resource scheduling, but also has significant theoretical value for achieving economical, safe, and green low-altitude operations. Summary of the Invention

[0003] The purpose of this invention is to propose a scheduling method for a low-altitude drone swarm including a single drone battery-swapping carrier, based on the transit search algorithm. This method significantly improves the scheduling efficiency and endurance of the drone swarm and effectively solves the problem of limited endurance of existing drones.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] This invention proposes a scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier escort, based on a transit search algorithm. The method includes the following steps:

[0006] Step S1: Obtain information on the drone cluster, mission information, and battery swapping carrier;

[0007] Step S2: Based on the operational tasks of the drone swarm, determine the scheduling decision variables for the drone swarm, including the accompanying flight of a single drone battery swapping carrier.

[0008] Step S3: Determine the scheduling objective function for the drone swarm including the single-drone battery-swapping carrier escort;

[0009] Step S4: Determine the constraints for scheduling a drone swarm including a single drone battery swapping carrier escort, thereby optimizing the scheduling objective function of the drone swarm;

[0010] Step S5: Based on the scheduling objective function and constraints of the UAV swarm, establish a scheduling model for the UAV swarm in low airspace with a single battery-swapping carrier accompanying it.

[0011] Step S6: Using the improved transit search algorithm based on machine learning, solve the scheduling model of the UAV swarm including the single UAV battery swapping carrier escort based on the UAV swarm scheduling decision variables, and obtain the scheduling scheme of UAVs and escort carrier.

[0012] Furthermore, the aforementioned drone swarm information includes drone ensembles. Drone battery capacity C, energy consumption per unit time when the drone speed is v e u (v) Maximum speed of the drone and minimum speed Drone battery swapping time Δt;

[0013] Task information includes a set of nodes. The set containing the starting point O, the ending point D, and the edges. And the side The distance is d ij Discrete time window set T = (t0, t...) 0+r ,t 0+2r …,t 0+mr ,…T max ), where r is the length of the time window and T is the latest completion time of the task. max ;

[0014] Information on the battery-swapping carrier includes the energy consumption e(v) per unit time when the carrier travels at a speed of v, and the carrier's maximum speed V. max .

[0015] Furthermore, the aforementioned drone swarm scheduling decision variables include drones. Has it passed through the edge? Defined as x k,i,j ∈{0,1}, drone Reaching the node Time t k,i drones At the node Is the battery replaced within the time window m?k,i,m ∈{0,1}, drone On the side Flight speed v k,i,j Is the battery-swapping carrier located at node m within time window m? And z m,i ∈{0,1}, the battery-swapping carrier is at node m in time window m. Is it a drone? Battery swapping k,i,m ∈{0,1}.

[0016] Furthermore, by analyzing the completion time and total energy consumption of the UAV swarm flight missions, the scheduling objective function for the UAV swarm-battery swapping carrier is determined, specifically as follows:

[0017]

[0018] in, This represents the latest arrival time of all drones at the destination, where α is the time weight and β is the energy consumption weight. This represents the total energy consumption of the drone swarm and the battery-swapping carrier, of which This indicates the flight energy consumption of a drone swarm. d represents the energy consumption of the battery-swapping carrier moving within a time window. prev(m,i) This represents the distance required to move from time window m to node i.

[0019] Furthermore, the constraints for the aforementioned drone swarm scheduling include path continuity constraints, time continuity constraints including battery swapping time, power constraints, time window matching constraints between the battery swapping carrier and the drone swarm, service capacity limit constraints of the battery swapping carrier, dwell time constraints of the battery swapping carrier, movement constraints of the battery swapping carrier, and mission time feasibility constraints.

[0020] Furthermore, the aforementioned path continuity constraint is used to ensure that the UAV completes the task from the starting point to the destination, specifically as follows:

[0021]

[0022] This means that each drone starts from the origin O and selects only one subsequent node;

[0023]

[0024] This indicates that each drone eventually reaches the destination D, and only one preceding node is selected for entry;

[0025]

[0026] This means that for intermediate nodes, the number of edges entering and leaving are equal, ensuring path continuity;

[0027] The time continuity constraint, including battery swapping time, is used to ensure the time sequence and the impact of battery swapping time, specifically:

[0028]

[0029] Among them, t k,j It is the time to reach node j, which must satisfy the flight time from i to j. And the possible battery swapping time Δt·∑ m y k,i,m ;

[0030]

[0031] M is a local constant, when x k,i,j =0 indicates a relaxed constraint, x k,i,j =1 ensures time continuity;

[0032] By setting the auxiliary variable Y k,i The power constraints are determined, and the auxiliary variables are:

[0033]

[0034] For the conditional judgment of calculating the amount of electricity, Y k,i =1 indicates that drone k swaps batteries at node i, then The battery level is reset to decrease after full charge; Y k,i =0 means no battery replacement, then Battery level decreased;

[0035] The drone swarm must maintain sufficient battery power to support its mission during flight, with the following specific constraints:

[0036]

[0037] If Y k,i =0, then If the power decreases, Y k,i =1, then The battery level is reset to decrease after full charge, and M ensures that the constraint only takes effect under the corresponding conditions.

[0038]

[0039] To ensure mission feasibility, a non-negative constraint on the drone's battery power is imposed.

[0040]

[0041] To limit the number of times a drone can have its battery swapped, each drone can only have its battery swapped once at the same node i to prevent duplicate battery swaps.

[0042] The specific time window matching constraints between the battery swapping carrier and the drone swarm are as follows:

[0043]

[0044] If y k,i,m =1, battery swapping time t k,i It must be completed within the time window [t0+m·r,t0+(m+1)·r]; M ensures that the constraint only takes effect during battery swapping;

[0045] The service capacity limit constraint for battery swapping motherships is used to limit the service capabilities of the mothership, specifically:

[0046]

[0047] This indicates that within the time window m of node i, the battery-swapping carrier can serve at most [number] services. Battery swapping;

[0048] The specific time constraints for the battery-swapping carrier's stay are as follows:

[0049]

[0050] This means that if the battery swapping carrier is located at node i in time window m, it must stay for at least time Δt to ensure that the battery swapping carrier has enough time to serve the drone.

[0051] The specific movement constraints of the battery-swapping carrier are as follows:

[0052]

[0053] in, This indicates that the battery-swapping carrier starts from the previous node i. prev The actual distance traveled to node i, V max ·(r-Δt) represents the maximum distance the battery-swapping carrier can move within a time window; For relaxation terms, The logical judgment value indicates the continuous stationing status of the battery swapping carrier, ensuring that the movement of the battery swapping carrier complies with the constraints of speed and time;

[0054] The specific time feasibility constraints for the task are as follows:

[0055]

[0056] Ensure that each drone k reaches the destination D within a time t k,D Not exceeding the maximum deadline T of the task max , t k,D T refers to the specific time it takes for drone k to reach its destination. max It is the latest time allowed to complete the task;

[0057]

[0058] Let v represent the flight speed of each drone k on edge (i,j). k,i,j The requirement is to achieve the minimum speed. and maximum speed between.

[0059] Furthermore, step S6 above specifically includes:

[0060] Step S61: Initial solution generation and fitness evaluation: Randomly generate diverse initial solutions. Each solution contains decision variables, such as path decision variables, time window allocation decisions, UAV speed decision variables, and UAV matching with the battery swapping carrier decision variables. Each decision variable must satisfy the corresponding constraints, and the fitness of each solution is calculated.

[0061] Step S62: Fusion optimization strategy: Based on the properties of the decision variables, different algorithms are used to fuse and optimize the decision variables in the solution vector;

[0062] Step S63: Search Phase: The optimization process of the solution is analogous to the dynamic adjustment mechanism of planetary orbital motion and luminosity changes. The fitness improvement of the solution is detected by the transit phase. If the transit phenomenon occurs, the planetary phase is used for global update; otherwise, the neighbor phase is used for local search update.

[0063] Step S64: Exploration Phase: Further expand the exploration range of the solution space by expanding the output solution population of the search phase, and further break the local convergence through multimodal perturbation;

[0064] Step S65: Iteration and Decoding: Calculate the fitness of the new solution population to obtain the optimal solution of the current solution population. Repeat the above operation steps until the maximum number of iterations is reached to obtain the solution vector with the lowest fitness value. Decode the optimal solution vector to obtain the scheduling scheme of UAVs and escort carriers.

[0065] The scheduling method for a low-altitude drone swarm including a single UAV battery swapping carrier accompanied by a carrier, based on the transit search algorithm described in this invention, can be entirely implemented using computer software. Therefore, correspondingly, this invention also provides a scheduling system for a low-altitude drone swarm including a single UAV battery swapping carrier accompanied by a carrier, based on the transit search algorithm. The system includes a storage device, which is used to execute the methods and steps described above.

[0066] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the scheduling method for a low-altitude UAV swarm including a single UAV battery swapping carrier accompanied by a carrier, based on the transit search algorithm described above.

[0067] The present invention also provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the scheduling method for a low-altitude UAV swarm with a single UAV battery swapping carrier escorted by a transit search algorithm as described in any one of the above-mentioned methods.

[0068] The beneficial effects of this invention are as follows:

[0069] This invention proposes a collaborative scheduling method for a carrier-UAV swarm based on an improved transit search algorithm. Through decoupling of spatiotemporal coupling constraints and a dynamic battery swapping decision mechanism, it achieves breakthrough improvements in endurance, environmental adaptability, and collaborative efficiency. Compared to traditional methods, it effectively solves the problem of UAV endurance limitations during long-duration missions, providing the possibility for UAV power replenishment. Simultaneously, this scheduling method exhibits superior adaptability and accuracy in dynamic and complex environments, optimizes resource utilization, improves overall mission execution efficiency, enhances the collaborative operation mechanism between UAVs and the carrier, and achieves Pareto optimality in energy consumption and efficiency. It provides a complete solution with both theoretical innovation and engineering practical value for large-scale UAV operation in low-altitude economic scenarios.

[0070] Furthermore, this invention balances the resource demands of the battery-swapping mothership and UAVs by setting time window matching, service capacity limits, and movement constraints, thereby reducing resource waste; it also forms a closed-loop guarantee mechanism by setting path, time, and power constraints to ensure mission integrity and safety; and it achieves global collaboration through constraint linkage between UAVs and the mothership (such as time window alignment and battery swapping decisions), thereby shortening the total mission time and reducing energy consumption.

[0071] This invention is applicable to the scheduling and energy management of unmanned aerial vehicle (UAV) swarms. Attached Figure Description

[0072] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0073] Figure 1 is a flowchart of a scheduling method for a low-altitude UAV swarm including a single UAV battery swapping carrier escorted by a transit search algorithm proposed in this invention.

[0074] Figure 2 is a flowchart of the scheduling scheme of UAVs and escorting motherships obtained by solving the scheduling model of a UAV swarm including a single UAV battery swapping mothership using the improved transit search algorithm based on machine learning described in this invention.

[0075] Figure 3 is a battery swapping decision and scheduling diagram for the 10 UAVs described in this invention.

[0076] Figure 4 is a battery swapping decision and scheduling diagram for the 20 UAVs described in this invention;

[0077] Figure 5 is a battery swapping decision-making and scheduling diagram for the 40 UAVs described in this invention.

[0078] Figure 6 is a performance comparison chart between the improved transit search algorithm proposed in this invention and existing algorithms. Detailed Implementation

[0079] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0080] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

[0081] Implementation Method 1: Referring to Figure 1, this implementation method proposes a scheduling method for a low-altitude UAV swarm with a single UAV battery swapping carrier escorted by a transit search algorithm. This method significantly improves the scheduling efficiency and endurance of the UAV swarm and effectively solves the problem of limited endurance of existing UAVs.

[0082] The scheduling method includes the following steps, as shown in Figure 1:

[0083] Step S1: Obtain information on the drone cluster, mission information, and battery swapping carrier;

[0084] Step S2: Based on the operational tasks of the drone swarm, determine the scheduling decision variables for the drone swarm, including the accompanying flight of a single drone battery swapping carrier.

[0085] Step S3: Determine the scheduling objective function for the drone swarm including the single-drone battery-swapping carrier escort;

[0086] Step S4: Determine the constraints for scheduling a drone swarm including a single drone battery swapping carrier escort, thereby optimizing the scheduling objective function of the drone swarm;

[0087] Step S5: Based on the scheduling objective function and constraints of the UAV swarm, establish a scheduling model for the UAV swarm in low airspace with a single battery-swapping carrier accompanying it.

[0088] Step S6: Using the improved transit search algorithm based on machine learning, solve the scheduling model of the UAV swarm including the single UAV battery swapping carrier escort based on the UAV swarm scheduling decision variables, and obtain the scheduling scheme of UAVs and escort carrier.

[0089] This embodiment proposes a collaborative scheduling method for a carrier-UAV swarm based on an improved transit search algorithm. Through decoupling of spatiotemporal coupling constraints and a dynamic battery swapping decision mechanism, it achieves breakthrough improvements in endurance, environmental adaptability, and collaborative efficiency. Compared to traditional methods, it effectively solves the problem of UAV endurance limitations during long-duration missions, providing the possibility for UAV power replenishment. Simultaneously, this scheduling method exhibits superior adaptability and accuracy in dynamic and complex environments, optimizes resource utilization, improves overall mission execution efficiency, enhances the collaborative operation mechanism between UAVs and the carrier, and achieves Pareto optimality in energy consumption and efficiency. It provides a complete solution with both theoretical innovation and engineering practical value for large-scale UAV operation in low-altitude economic scenarios.

[0090] Implementation Method 2: Refer to Figures 2 to 5 for an explanation of this implementation method. This implementation method is a detailed explanation of the scheduling method for a low-altitude UAV swarm with a single UAV battery swapping carrier escorted by a transit search algorithm proposed in Implementation Method 1 above.

[0091] Step S1: Obtain information on the drone cluster, mission information, and battery swapping carrier;

[0092] Specifically:

[0093] Drone swarm information includes drone ensembles Drone battery capacity C, energy consumption per unit time when the drone speed is v e u (v) Maximum speed of the drone and minimum speed Drone battery swapping time Δt;

[0094] Task information includes a set of nodes. The set containing the starting point O, the ending point D, and the edges. And the side The distance is d ij Discrete time window set T = (t0, t...) 0+r ,t 0+2r …,t 0+mr ,…T max ), where r is the length of the time window and T is the latest completion time of the task. max ;

[0095] Information on the battery-swapping carrier includes the energy consumption e(v) per unit time when the carrier travels at a speed of v, and the carrier's maximum speed V. max .

[0096] Step S2: Based on the operational tasks of the drone swarm, determine the scheduling decision variables for the drone swarm, including the accompanying flight of a single drone battery swapping carrier.

[0097] Specifically:

[0098] The decision variables for drone swarm scheduling include drones Has it passed through the edge? Defined as x k,i,j ∈{0,1}, drone Reaching the node Time t k,i drones At the node Is the battery replaced within the time window m? k,i,m ∈{0,1}, drone On the side Flight speed v k,i,j Is the battery-swapping carrier located at node m within time window m? And z m,i ∈{0,1}, the battery-swapping carrier is at node m in time window m. Is it a drone? Battery swapping k,i,m ∈{0,1}.

[0099] Step S3: Determine the scheduling objective function for the drone swarm including the single-drone battery-swapping carrier escort;

[0100] Specifically:

[0101] Considering the completion time T of the drone swarm flight missionmax and total energy consumption E totla Set the synthesis objective function:

[0102]

[0103] Where, max k∈K (t k,D -t0) represents the latest time that all drones arrive at the destination (mission completion time), and α is the time weight value. This represents the total energy consumption of the drone swarm, consisting of the flight energy consumption of the drone swarm and the movement energy consumption of the battery-swapping carrier. β is the energy consumption weighting value. Specifically, This represents the flight energy consumption of a drone swarm, which is the sum of the energy consumption of each drone on each segment of the path (energy consumption per unit time at speed v × flight time). d represents the energy consumption of the battery-swapping carrier moving within a time window. prev(m,i) This represents the distance required to move from time window m to node l.

[0104] Step S4: Determine the constraints for scheduling a drone swarm including a single drone battery swapping carrier escort, thereby optimizing the scheduling objective function of the drone swarm;

[0105] Specifically:

[0106] To optimize the overall objective function, the scheduling of a drone swarm, including single-UAV battery swapping carrier escort, must satisfy the following constraints;

[0107] Step S4.1 Path continuity constraint: Used to ensure that the UAV completes the task from the starting point to the destination.

[0108]

[0109] This means that each drone starts from the starting point O and selects only one subsequent node.

[0110]

[0111] This indicates that each drone eventually reaches the destination D, and only one preceding node is selected for entry.

[0112]

[0113] This means that for intermediate nodes, the number of edges entering and leaving are equal, ensuring path continuity.

[0114] This implementation sets path continuity constraints to ensure the continuity of the UAV's path from start to finish, avoiding path interruptions or repeated visits to nodes. The effect is to reduce invalid paths: preventing UAVs from getting stuck in infinite loops or missing critical nodes during the mission, directly reducing total flight distance and energy consumption. Mission integrity: ensuring all UAVs complete the mission from start to finish.

[0115] Step S4.2 includes time continuity constraints on battery swapping time: ensuring the time sequence and the impact of battery swapping time.

[0116]

[0117] Among them, t k,j It is the time to reach node j, which must satisfy the flight time from i to j. And the possible battery swapping time Δt·∑ m y k,i,m .

[0118]

[0119] M is a local constant, when x k,i,j =0 indicates a relaxed constraint, x k,i,j =1 ensures continuous time.

[0120] This implementation sets time continuity constraints to ensure logical consistency of time variables by introducing the influence of time sequence and battery swapping time. The effect is to avoid time conflicts: by using time window constraints, the time sequence of the UAV during battery swapping and flight is guaranteed to be reasonable, reducing mission delays. Precise scheduling: by combining battery swapping time, the allocation of time resources is optimized, supporting optimization objectives related to total mission time or timeliness in the objective function.

[0121] Step S4.3 Power Constraint:

[0122] Determined by auxiliary variable Y k,i :

[0123]

[0124] For the conditional judgment of calculating the amount of electricity, Y k,i =1 indicates that drone k swaps batteries at node i, then The battery level is reset to decrease after full charge; Y k,i =0 means no battery replacement, then Battery level is decreasing.

[0125] The drone swarm must maintain sufficient battery power to support its mission during flight, with the following specific constraints:

[0126]

[0127] If Y k,i =0, then Battery level decreases. If Y... k,i =1, then The battery level is reset to decrease after full charge. M ensures that the constraint only takes effect under the corresponding conditions.

[0128]

[0129] The drone's battery level is constrained to be non-negative, ensuring mission feasibility.

[0130]

[0131] To limit the number of times a drone can have its battery swapped, each drone can only have its battery swapped once at the same node i, in order to prevent duplicate battery swaps.

[0132] This implementation sets power constraints to manage the drone's power consumption and battery swapping strategy, ensuring that the power supply always meets flight requirements. The effect is to prevent power depletion: through battery swapping and consumption calculations, it avoids mission failure due to insufficient power, improving system reliability. Dynamic battery swapping decision-making: through auxiliary variable Y... k,i Differentiate between battery swapping and non-battery swapping scenarios, optimize battery swapping location selection strategies, ensure orderly battery swapping for drone swarms, and improve system coordination.

[0133] Step S4.4 Time window matching constraints between the battery swapping carrier and the UAV swarm:

[0134]

[0135] If y k,o,m =1, battery swapping time t k,i It must be completed within the time window [t0+m·r,t0+(m+1)·r]; M ensures that the constraint only takes effect during battery swapping.

[0136] This implementation sets a time window matching constraint for the battery-swapping mothership, aiming to align the UAV's battery-swapping time with the mothership's time window. The benefits include: efficient resource utilization: ensuring that UAVs only swap batteries within the mothership's available time window, achieving time matching between UAVs and the battery-swapping mothership during dynamic flight. Cooperative scheduling: supporting orderly battery swapping of multiple UAVs within the time window, shortening the total waiting time, and directly optimizing time-related indicators in the objective function.

[0137] Step S4.5 Battery Swapping Carrier Service Capacity Limitation Constraint: Limits the service capacity of the carrier.

[0138]

[0139] This indicates that within the time window m of node i, the battery-swapping carrier can serve at most [number] services. Battery swap.

[0140] This implementation sets a service capacity limit constraint on the mothership, which limits the maximum number of battery swaps the mothership can perform within a single time window. The effect is load balancing: preventing the mothership from being overloaded during a certain period and reducing the risk of service delays.

[0141] Step S4.6 Battery swapping carrier dwell time constraint:

[0142]

[0143] This means that if the battery swapping carrier is located at node i in time window m, it must stay for at least time Δt to ensure that the battery swapping carrier has enough time to serve the drone.

[0144] This implementation sets a constraint on the mothership's dwell time, requiring it to remain at the node for a sufficient period to complete the battery swapping service. The effect is improved service reliability: ensuring the integrity of the battery swapping operation and preventing failure due to the mothership leaving too early.

[0145] Step S4.7 Battery swapping carrier movement constraints:

[0146]

[0147] in, This indicates that the battery-swapping carrier starts from the previous node i. prev The actual distance traveled to node i, V max ·(r-Δt) represents the maximum distance the battery-swapping carrier can move within a time window; For relaxation terms, This represents the logical judgment value indicating the continuous stationary state of the battery swapping carrier, ensuring that the movement of the battery swapping carrier complies with speed and time constraints.

[0148] This implementation sets movement constraints for the mothership to limit its movement distance between different time windows. The benefits are: Safety assurance: By limiting the mothership's speed, the movement distance is restricted, preventing accidents caused by the mothership exceeding a speed threshold. Time window coherence: This ensures the mothership's reachability in adjacent time windows, preventing service interruptions due to exceeding movement limits.

[0149] Step S4.8 Task time feasibility constraints:

[0150]

[0151] Ensure that each drone k reaches the destination D within a time t k,D Not exceeding the maximum deadline T of the task max , t k,D T refers to the specific time it takes for drone k to reach its destination. max It is the latest time allowed to complete the task.

[0152]

[0153] Let v represent the flight speed of each drone k on edge (i,j). k,i,j The requirement is to achieve the minimum speed. and maximum speed between.

[0154] Step S5: Based on the scheduling objective function and constraints of the UAV swarm, establish a scheduling model for a single battery-swapping carrier escorting UAV swarm in low-altitude airspace.

[0155] Step S6: Use the improved transit search algorithm based on machine learning to solve the scheduling model of the UAV swarm including the single UAV battery swapping carrier escort, and obtain the scheduling scheme of UAVs and escort carrier.

[0156] Specifically:

[0157] As shown in Figure 2:

[0158] Step S61: Initial solution generation and fitness evaluation: Randomly generate diverse initial solutions. Each solution contains decision variables, such as path decision variables, time window allocation decisions, UAV speed decision variables, and UAV matching with the battery swapping carrier decision variables. Each decision variable must satisfy the corresponding constraints, and the fitness of each solution is calculated.

[0159] Step S62: Fusion optimization strategy: Based on the properties of the decision variables, different algorithms are used to fuse and optimize the decision variables in the solution vector;

[0160] Step S63: Search Phase: The optimization process of the solution is analogous to the dynamic adjustment mechanism of planetary orbital motion and luminosity changes. The fitness improvement of the solution is detected by the transit phase. If the transit phenomenon occurs, the planetary phase is used for global update; otherwise, the neighbor phase is used for local search update.

[0161] Step S64: Exploration Phase: Further expand the exploration range of the solution space by expanding the output solution population of the search phase, and further break the local convergence through multimodal perturbation;

[0162] Step S65: Iteration and Decoding: Calculate the fitness of the new solution population to obtain the optimal solution of the current solution population. Repeat the above operation steps until the maximum number of iterations is reached to obtain the solution vector with the lowest fitness value. Decode the optimal solution vector to obtain the scheduling scheme of UAVs and escort carriers.

[0163] More specifically:

[0164] A1 Galaxy Phase: Initial Solution Generation and Fitness Assessment.

[0165] A1.1 Initial solution generation generates diverse initial solutions that satisfy the constraints, laying the foundation for global search.

[0166] Randomly generate ns initial solutions, and the set of solutions is defined as S = {S1, S2, ..., S...} ns}, where each solution S p Includes decision variables p is the index of the solution.

[0167] Path decision variables The generation satisfies the following constraints:

[0168]

[0169] in, This represents the probability that drone k chooses edge (i,j) for navigation, along with the distance d. ij The probability is inversely proportional to the probability of node i; in addition, the sum of the probabilities from node i to all possible subsequent nodes j′ is 1 to ensure the continuity of the UAV's path.

[0170] Time window allocation decision variables Generation:

[0171]

[0172] in, Let represent the time when drone k arrives at node i, which follows a uniform distribution defined on the interval (t0+m·r, t0+(m+1)·r-Δt). Random sampling ensures unbiased exploration of the search space.

[0173] Speed ​​decision variables The initialization satisfies the following conditions:

[0174]

[0175] in, Let the speed of drone k on edge (i,j) conform to the definition of speed within the interval A uniform distribution on the surface.

[0176] Decision variables for matching drones with battery-swapping carriers The generation satisfies the following conditions:

[0177]

[0178] in, This indicates whether drone k swaps batteries within node i and time window m. If it swaps batteries, the value is [value to be filled in]. =1, and simultaneously satisfy Otherwise The value is 0, and at the same time Given random probability p swap and arrival time Whether to generate randomly within a time window, where p swap For battery swapping probability, The time when drone k arrives at node i is represented by t0, rand() is a random number generator used to generate uniform random numbers in [0,1], t0+m·r is the start time of time window m, and t0+(m+1)·r-Δt is the end time of time window m minus the power swapping time.

[0179] A1.2 Fitness Evaluation of Solutions: Evaluate the quality of each solution to guide subsequent optimization.

[0180] Calculate the fitness F(S) of each solution in set S. p ):

[0181]

[0182] Among them, F(S) p The sum of time and energy costs for completing the task is denoted as α, where α and β are trade-off coefficients. Furthermore, a penalty term is introduced to eliminate infeasible solutions.

[0183]

[0184] Where μ is the penalty coefficient, when the electricity... or time In such cases, penalty clauses are added to ensure the reasonableness of the solution.

[0185] A2 fusion optimization strategy.

[0186] Input population S, output population S′.

[0187] A fusion optimization is performed on the existing solution set S to improve the quality of the solutions and reduce unnecessary space exploration. Based on the properties of the decision variables, different algorithms are used to fuse and optimize the decision variables in the solution vector as follows:

[0188] A2.1 improves path and velocity decision variables through cross-referencing and fusion, namely:

[0189] Cross-optimization of path decision variables.

[0190] The optimal solution S in the solution set S is obtained based on A1. * Introduce crossover probabilities and update each solution S in the solution set S. p Path decision variables:

[0191]

[0192] in, This indicates a new path decision change for drone k.

[0193] The quantity, rand() is a random function used to generate uniformly random numbers in the range [0,1]. c It's the crossover probability. The optimal solution S * The path decision variables of UAV k The current solution S p Path decision-making for UAV k in China.

[0194] Optimization of speed decision variables through fusion.

[0195] The optimal solution S in the solution set S is obtained based on A1. * Introduce a fusion factor to update each solution S in the solution set S. p Speed ​​decision variables:

[0196]

[0197] in, This represents the new flight speed of drone k on the flight path (i,j). Optimal solution S * The flight speed of the UAV k The current solution S p The flight speed of the UAV k is given by δ, which is a fusion factor.

[0198] A2.2 Local optimization of battery swapping time window based on LSTM prediction improvement.

[0199] Considering the discreteness of the battery swapping time window, and fully leveraging the spatial search efficiency and accuracy of machine learning algorithms, the LSTM algorithm is further introduced to optimize and improve the battery swapping time window:

[0200]

[0201] Where, m new This represents the optimal battery swapping time window for drone k at node i in the new solution. LSTM() is a Long Short-Term Memory network that maps input features (time, battery level, distance) to a time window index to output predicted values. Let k be the time it takes for drone k to reach node l. This is the drone's current battery level. This represents the total remaining path distance. LSTM is based on the current time. Current battery level The total distance of the remaining path is used to predict the optimal battery swapping time window m. new .

[0202] A3. Search phase.

[0203] The optimization process of the solution is analogous to the dynamic adjustment mechanism of planetary orbital motion and luminosity changes. The transit stage (A3.1) is used to detect the fitness improvement of the solution (i.e., to refine the population S′ from A2). If a transit occurs, a global update is performed using the planetary stage (A3.2); otherwise, a local search update is performed using the neighbor stage (A3.3).

[0204] A3.1 Transit phase.

[0205] For the candidate solution population S′={S′1,S′2,…,S′ ns Fitness assessment is performed by calculating the relative quality of the solution using photometric methods, such as L. b :

[0206]

[0207] Where, d b The solution S′ is represented p The distance to the current optimal solution. For each solution vector S′ in the solution population S′. p Calculate the initial photometric value Then for S′ p By performing a small random perturbation, a new corresponding solution vector S″ is obtained. p Recalculate photometric values like If the transit occurs, the planetary phase begins; otherwise, the neighbor phase begins.

[0208] A3.2 Planetary Phase.

[0209] The solution vector S″ for detecting the transit p Perform global refinement to simulate the orbital adjustments of planets influenced by the gravitational pull of stars. The current optimal solution S is then... * With disturbance solution S″ pWeighted fusion to generate S z As shown below:

[0210]

[0211] In the formula, c1 is the gravitational balance coefficient, which controls the contribution weight of the historical optimal solution; R L To simulate the uncertainty of celestial motion using stochastic perturbation forces, based on the galaxy center S... r (initial population optimal solution S) 0 Or the current optimal solution S * Further adjustments to the specific solution vector are shown in the following formula:

[0212]

[0213] Where g (1,2,3) represents the random selection of perturbation mode; c2 and c3 represent perturbation amplitude coefficients, which determine the search range.

[0214] A3.3 Neighbor Phase.

[0215] For the perturbation solution S″ where no transit was detected p Fine-tuning is performed to simulate local orbital fine-tuning driven by stellar light pressure, combined with the perturbation solution S″. p With the galaxy center S r The intermediate solution is generated as shown in the following formula:

[0216]

[0217] In the formula, c4 is the centroid offset coefficient, which controls the conservatism of the local search. Furthermore, the corresponding solution vector is further adjusted based on the intermediate solution and the galactic center, as shown in the following formula:

[0218]

[0219] Where c5 and c6 are the local perturbation step sizes. This is achieved by adjusting the initial solution population S′={S′1,S′2,…,S′...}. ns After performing transit detection and perturbation update, output the new solution population S″′={S″′1,S″′2,…,S″′ ns}

[0220] A4. Exploration Phase.

[0221] Finally, the solution population S″' output during the search phase is further expanded to broaden the exploration range of the solution space. Multimodal perturbation is used to further break local convergence and enhance the traversal capability of the solution space. This is based on a solution vector S″' output by A3. pAt the galactic center, multiple random perturbation terms are introduced to calculate S″′. p As shown in the formula below:

[0222]

[0223] Where c7 is the stellar gravitational decay coefficient, controlling the proportion of historical solutions retained; c8 and c9 are perturbation factors, with c8 being the baseline intensity and c9 being the distance decay factor. P~U(0,1) is the random perturbation factor. h~{1,2,3,4} is the condition variable. The solution population S″′={S″′1,S″′2,…,S″′ is generated during the search phase. ns After applying multimodal perturbation, a new solution population S””={S″″1,S″″2,…,S″″} is generated. ns}

[0224] A5. Calculate the fitness of the new solution population according to step A1.2 above, and obtain the optimal solution minF(S””) of the current solution population. Repeat steps A2-A5 until the maximum number of iterations maxgen is reached, and obtain the solution vector with the lowest fitness value. Decode the optimal solution vector to obtain the scheduling scheme of UAV and escort carrier, as shown in Figures 3 to 5.

[0225] Implementation Method 3: The scheduling method for a low-altitude UAV swarm with a single UAV battery-swapping carrier escorted by a transit search algorithm proposed in the above implementation methods can be entirely implemented using computer software. Therefore, correspondingly, this implementation method proposes a scheduling system for a low-altitude UAV swarm with a single UAV battery-swapping carrier escorted by a transit search algorithm. The system includes:

[0226] Storage device used to acquire information on drone swarms, mission information, and battery swapping carriers;

[0227] A storage device for determining the scheduling decision variables of a drone swarm, including the escort of a single drone battery-swapping carrier, based on the drone swarm's operational tasks.

[0228] A storage device for determining the scheduling objective function of a drone swarm that includes a single drone battery-swapping carrier escorting it;

[0229] A storage device for determining the constraints of drone swarm scheduling, including single drone battery swapping carrier escort, thereby optimizing the drone swarm scheduling objective function;

[0230] A storage device for establishing a scheduling model of a UAV swarm in low airspace with a single battery-swapping carrier accompanying it, based on the scheduling objective function and constraints of the UAV swarm.

[0231] A storage device for using a transit search algorithm based on machine learning to solve the scheduling model of a drone swarm including a single drone battery swapping carrier escort, based on the drone swarm scheduling decision variables, and to obtain the scheduling scheme of drones and escort carriers.

[0232] Implementation Method 4: This implementation method provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it executes the scheduling method for a low-altitude UAV swarm with a single UAV battery swapping carrier accompanied by a carrier, based on the transit search algorithm described in any of the above implementation methods.

[0233] Implementation Method 5: This implementation method provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the scheduling method for a low-altitude UAV swarm with a single UAV battery swapping carrier escorted by a transit search algorithm, as described in any of the above implementation methods.

[0234] This embodiment provides a computer device, the hardware of which is a general-purpose model and is not shown in the figure. The system includes a processor and a memory, which can be connected by a bus or other means. The memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs and modules, as well as corresponding program instructions / modules. The processor executes various functional applications and data processing by running the non-transitory software programs, instructions and modules stored in the memory, so as to realize the scheduling method and steps of the UAV swarm with single UAV battery swapping carrier escort based on the transit search algorithm in the above method embodiment.

[0235] Implementation Method Six: Refer to Figure 6 for an explanation of this implementation method. This implementation method compares the improved transit search algorithm based on machine learning proposed in the above implementation methods with existing algorithms.

[0236] By setting different numbers of drone swarms, the transit search algorithm based on machine learning was compared with the genetic algorithm and the simulated annealing algorithm. The comparison indicators were the quality of the solution (average fitness), the convergence time (in seconds), and the number of iterations (average).

[0237] The calculation results are shown in Figure 6. As can be seen from the figure, with 10 drones, the improved transit search algorithm has an average solution value of 38,186,296.57, a convergence time of 7.24, and 38.10 iterations. Compared to the genetic algorithm and simulated annealing algorithm, it is optimal in terms of solution quality, convergence time, and number of iterations. Furthermore, with an increased number of drones of 20 and 40, it can be seen that as the number of drones increases, the improved transit search algorithm still outperforms the genetic algorithm and simulated annealing algorithm in terms of solution quality, convergence time, and number of iterations.

[0238] Those skilled in the art will understand that the above description is merely a preferred embodiment of the present invention, and the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. This is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0239] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier escorted by a transit search algorithm, characterized in that, The method is as follows: S1: Obtain UAV swarm information, mission information, and battery swapping carrier information; S2: Determine the scheduling decision variables for the UAV swarm, including single UAVs accompanied by battery swapping carriers, based on the UAV swarm's operational mission; S3: Determine the scheduling objective function for the UAV swarm, including single UAVs accompanied by battery swapping carriers; specifically: determine the scheduling objective function for the UAV swarm-battery swapping carrier based on the completion time and total energy consumption of the UAV swarm's flight mission, expressed as: in, This indicates the latest time that all drones will arrive at the destination. It is a time-weighted value. It is an energy consumption weighting value. This indicates the flight energy consumption of a drone swarm. This indicates the energy consumption of the battery-swapping carrier as it moves within a time window. Indicates time window Move to node S4: Determine the constraints for scheduling a drone swarm including a single drone battery-swapping carrier, thereby optimizing the scheduling objective function of the drone swarm; S5: Based on the scheduling objective function and constraints of the drone swarm, establish a low-altitude drone swarm scheduling model including a single drone battery-swapping carrier; S6: Use a transit search algorithm based on machine learning to solve the scheduling model of the drone swarm including a single drone battery-swapping carrier based on the drone swarm scheduling decision variables, and obtain the scheduling scheme of drones and carriers; The improved transit search algorithm includes a search phase, which compares the optimization process of the solution to the dynamic adjustment mechanism of planetary orbital motion and light intensity changes. The transit phase is used to detect the fitness improvement of the solution. If a transit occurs, a planetary phase is used for global update; otherwise, a neighbor phase is used for local search update.

2. The scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier accompanied by a carrier, based on the transit search algorithm according to claim 1, is characterized in that, Drone swarm information includes drone ensembles The drone's battery capacity (C) and its flight speed are as follows: Energy consumption per unit time Maximum speed of drones and minimum speed Drone battery swapping time Task information includes a set of nodes. , including the starting point and the end point set of edges And the side The distance is Discrete-time window set ,and The time window length and the latest task completion time Information on the battery-swapping carrier includes its speed. Energy consumption per unit time e Maximum speed of the battery-swapping carrier 。 3. The scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier accompanied by a carrier, based on the transit search algorithm according to claim 2, is characterized in that... The decision variables for drone swarm scheduling include drones Has it passed through the edge? drones Reaching the node Time drones At the node Time window Should the battery be replaced? drones On the side flight speed Battery swapping carrier in time window Is it located at node? Battery swapping carrier during the time window At node For drones Battery swapping 。 4. The scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier accompanied by a carrier, based on the transit search algorithm according to claim 1, is characterized in that, The constraints for UAV swarm scheduling include path continuity constraints, time continuity constraints including battery swapping time, power constraints, time window matching constraints between the battery swapping carrier and the UAV swarm, service capacity limit constraints of the battery swapping carrier, dwell time constraints of the battery swapping carrier, movement constraints of the battery swapping carrier, and mission time feasibility constraints.

5. A scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier accompanied by a carrier, based on a transit search algorithm according to claim 4, characterized in that, Path continuity constraints: Time continuity constraints including battery swapping consumption: in, Is the node reached Time, From arrive Flight time, For battery swapping time, It is a maximum constant; by setting auxiliary variables Determine the power constraints and auxiliary variables: The specific power constraints for drone swarms during flight missions are as follows: Time window matching constraints between battery swapping carrier and drone swarm: Battery swapping carrier service capacity constraints: Battery swapping carrier dwell time constraints: Battery swapping carrier movement constraints: in, This indicates that the battery-swapping carrier has moved from the previous node. To the node The actual distance traveled This indicates the maximum distance the battery-swapping carrier can move within a time window; For relaxation terms, The logical judgment value representing the continuous stationary state of the battery swapping carrier ensures that the carrier's movement complies with speed and time constraints; mission time feasibility constraints: 。 6. A scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier accompanied by a carrier, based on a transit search algorithm according to claim 5, characterized in that, S6 specifically includes: S61: Initial solution generation and fitness evaluation: Randomly generate diverse initial solutions, each solution containing decision variables, such as path decision variables, time window allocation decisions, UAV speed decision variables, and UAV-battery swapping carrier matching decision variables. Each decision variable must satisfy the corresponding constraints, and the fitness of each solution is calculated. S62: Fusion Optimization Strategy: Based on the properties of the decision variables, different algorithms are used to fuse and optimize the decision variables in the solution vector. S63: Search Phase: The optimization process of the solution is analogous to the dynamic adjustment mechanism of planetary orbital motion and light intensity changes. The fitness improvement of the solution is detected using the transit phase. If a transit occurs, a global update is performed using the planetary phase; otherwise, a local search update is performed using the neighbor phase. S64: Exploration Phase: The exploration range of the solution space is further broadened for the output solution population of the search phase, and local convergence is further broken through multimodal perturbation. S65: Iteration and Decoding: The fitness of the new solution population is calculated to obtain the optimal solution of the current solution population. The above steps are repeated until the maximum number of iterations is reached, obtaining the solution vector with the lowest fitness value. The optimal solution vector is decoded to obtain the scheduling scheme of the UAV and the escort carrier.

7. A scheduling system for a low-altitude UAV swarm including a single UAV battery-swapping carrier escorted by a transit search algorithm, characterized in that, The system includes a storage device for performing the method and steps of claim 1.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the scheduling method for a low-altitude UAV swarm including a single UAV battery-swapping carrier, based on any one of claims 1-6, according to the transit search algorithm.

9. A computer device, characterized in that, The device includes a memory and a processor. The memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the scheduling method for a low-altitude UAV swarm with a single UAV battery swapping carrier accompanied by a carrier, based on any one of claims 1-6.

Images