Multi-target shortest flight path planning method for unmanned aerial vehicle

Abstract

The invention belongs to the technical field of unmanned aerial vehicle path planning, and particularly discloses an unmanned aerial vehicle multi-target shortest flight path planning method, which comprises the following steps: acquiring coordinate information of the maximum endurance mileage, a take-off point, a landing point, a task target point and a charging station of an unmanned aerial vehicle, and generating a distance matrix containing distances among all stations; generating an initial flight route based on a simulated annealing algorithm, generating a new state by randomly exchanging a target position in the route, and gradually reducing the temperature until convergence to obtain a single calculation optimal solution; independently operating the algorithm for at least three times, and determining a global optimal solution through a two-out-of-three comparison mechanism; and outputting an optimal solution containing the path and the mileage. According to the method, the accuracy of an optimal result is greatly improved through two-out-of-three comparison, the problem of uncertainty of a probabilistic algorithm is solved, meanwhile, a take-off and landing point is fixed, a charging station access rule is included, a target is forced to be accessed only once, and the adaptability and preciseness of path planning are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle (UAV) path planning, and in particular relates to a method for planning the shortest flight route for multiple targets of an UAV. Background Art

[0002] In the field of drone path planning technology, multi-objective shortest flight route planning is a core issue for improving mission efficiency and reliability. With the large-scale application of drones in scenarios such as logistics and distribution, inspection and monitoring, and emergency rescue, how to quickly generate optimal routes that meet flight range and site constraints in complex geographical environments has become a pressing technical challenge.

[0003] Traditional simulated annealing algorithms rely on the probabilistic acceptance mechanism of the Metropolis criterion. Although this algorithm can avoid local optimality to a certain extent, the randomness of single calculation results makes it impossible to meet the stringent reliability requirements of drone missions. Existing algorithms also fail to fully consider constraints such as fixed takeoff / landing points, limited flight time, and unique target access. The generated path may include illegal jumps or exceed the drone's flight range, rendering the planning result infeasible. Summary of the Invention

[0004] The purpose of the present invention is to overcome the defects in the prior art and provide a method for planning the shortest flight path for a multi-target unmanned aerial vehicle.

[0005] The present invention provides a method for planning the shortest flight path of a multi-objective unmanned aerial vehicle, comprising the following steps:

[0006] (1) Obtain the coordinate information of the maximum range of the UAV, take-off point, landing point, mission target point, and charging station, and generate a distance matrix containing the distances between all stations;

[0007] (2) Generate an initial flight path based on the simulated annealing algorithm, generate a new state by randomly exchanging the target position in the path, and gradually reduce the temperature until convergence to obtain a single calculation optimization solution;

[0008] (3) Run the simulated annealing algorithm at least three times independently to generate multiple single-calculation optimization solutions. If any two solutions are consistent and are the current optimal solutions, they are determined to be the global optimal solution. If they are inconsistent, only the current optimal solution is retained and the simulated annealing algorithm is rerun until any two solutions are consistent and are the optimal solutions.

[0009] (4) Output the optimal solution: including flight path and total mileage information.

[0010] A further solution is that the take-off point of the optimal solution flight route is the route starting point, and the landing point is the route ending point; each mission target is visited only once; and the flight distance between any adjacent stations does not exceed the current remaining range of the drone.

[0011] A further solution is that the initial flight route generation process is as follows: fix the positions of the take-off point and the landing point, randomly arrange the mission target points, and form the initial flight route L O ;

[0012] Calculate the total flight distance E of the initial flight route O , and set the current flight route L c =L O 、Current total flight distance E c =E O , and record the current optimal route L b =L c and the optimal total flight distance E b =E c .

[0013] A further solution is that in the simulated annealing algorithm, the initial temperature T 初始 =100, crystallization temperature T 结晶 =0;

[0014] Temperature attenuation coefficient α = 0.01;

[0015] Define the number of iterative optimizations under isothermal state C = 30 × i, the state stability counter threshold JS = 30, where i is the number of mission target points, and the judgment logic of the state stability counter threshold (JS = 30) is: if the optimal total flight distance E is not updated after 30 consecutive iterative optimizations b , the iterative optimization at the current temperature is terminated early.

[0016] A further solution is that, in the simulated annealing algorithm, the temperature decay process is:

[0017] Reduce the temperature according to the attenuation coefficient α: T 新 =T 当前 ×(1-α)

[0018] If the current temperature T 当前 >T 结晶 , a new round of iterative optimization is carried out; otherwise, the algorithm is terminated;

[0019] Among them, T 当前 Indicates the temperature value that the algorithm is using in a certain round of iteration, T 新 Represents the updated temperature value after temperature decay, that is, the temperature to be used in the next iteration.

[0020] A further solution is that the iterative optimization process of the simulated annealing algorithm under isothermal conditions is:

[0021] Random exchange node: Generate two different random numbers X and Y, and exchange the current flight path L cThe task points at the X and Y positions in the , generate a new route L' c ;

[0022] Calculate the total flight distance E' of the new route c , and calculate the distance difference ΔE=E' c -E b ;

[0023] Accept or reject the new path L' according to the Metropolis criterion c .

[0024] A further solution is that the Metropolis criterion is:

[0025] If ΔE<0, directly accept the new route and update L b =L' c and E b =E' c ;

[0026] If ΔE ≥ 0 (the new path is worse), accept the new path L' with probability P(ΔE) = exp(-ΔE / (kT)) c , where k is a constant and T is the current temperature.

[0027] A further solution is that the state stability judgment process of the simulated annealing algorithm is:

[0028] Repeat the iterative optimization and Metropolis criterion judgment until the number of isothermal iterations C at the current temperature is completed, or no better solution is found after JS consecutive iterations.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] This method verifies the optimal solution by comparing two out of three independent simulated annealing results. The method requires that at least two of the three results agree before the optimal solution is considered. If any two of the three results agree, the consensus solution is directly output; if not, the discrepant solution is discarded and regenerated until the matching criteria are met. This significantly improves the accuracy of the optimal solution, completely resolves the uncertainty problem of probabilistic algorithms, and ensures the absolute reliability of the output results.

[0031] This method eliminates invalid paths by specifying fixed takeoff and landing points as endpoints. It also incorporates unlimited access to charging stations, allowing for dynamic insertion of charging station nodes into the path, ensuring the drone completes its mission within its flight capabilities. This overcomes the drawback of traditional algorithms that fail to consider flight range limitations. By forcing each target point to be visited only once and combining this with the precise calculation of flight distances between stations in the distance matrix, the logical rigor of path planning is enhanced.

[0032] The present invention realizes a dynamic temperature control process by setting the initial temperature to a higher value, slowly cooling and crystallizing the temperature: in the high temperature stage, the probability acceptance mechanism is used to accept poor solutions with a high probability, thereby enhancing the global search capability; in the low temperature stage, the algorithm gradually converges to a single calculation optimization solution, thereby avoiding non-optimal solutions caused by rapid cooling. Combined with the number of isothermal iterations and the state stability counter threshold, the current temperature cycle is terminated in advance when the solution is not improved after multiple consecutive iterations, thereby reducing invalid calculations and improving computational efficiency. By randomly exchanging the order of target points in the path and combining the probability acceptance mechanism, the algorithm can not only escape the local optimal trap, but also quickly lock the high-quality solution through the stable counter in the low temperature stage, thereby achieving efficient optimization in complex multi-objective scenarios.

[0033] The output of this invention includes a complete flight route sequence, detailed distance information for each segment, and the total mileage. The format is highly standardized and can be directly integrated into drone mission management systems, reducing manual analysis costs. By using station coordinates, drone endurance, and the number of targets as input parameters, the system can adapt to mission scenarios of varying scales and meet diverse engineering needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The following drawings are merely provided for illustrative purposes only and are not intended to limit the scope of the present invention.

[0035] Figure 1 : Route planning flow chart of the present invention;

[0036] Figure 2 : Logic flow chart of simulated annealing algorithm. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution, design method and advantages of the present invention more clear, the present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0038] The present invention provides a method for planning the shortest flight path of a multi-target UAV. Figure 1 As shown, the following steps are included:

[0039] S1. Obtain the coordinate information of the maximum range of the UAV, take-off point, landing point, mission target point, and charging station, build a mathematical model, and generate a distance matrix containing the distances between all stations; the process of the mathematical model is as follows:

[0040] (1) Let n represent the number of mission target points, let k represent the number of charging stations, Represents the task target point i, a t Indicates the take-off station, a l Indicates the landing site, Indicates a charging station.

[0041] (2) Flight routes: Define a set Excluding charging stations, all elements in set A are a t As the first element, with a l Any permutation of the last element is defined as a flight path, and the number of possible flight paths is m!;

[0042] (3) Flight distance: The sum of the distances between all adjacent points on the flight path is the flight distance. Define d0 as the takeoff point a t The flight distance to the first target, defined as d m Indicates the flight distance from the last target to the landing point, d i represents the flight distance from the i-th target to the (i+1)-th target, and D represents the total flight distance.

[0043] (4) The mathematical representation of the algorithm is to find Established An arrangement of elements.

[0044] The process of generating a distance matrix is:

[0045] (1) Clarify the scope of the site;

[0046] Site type: including take-off point, landing point, mission target point and charging station;

[0047] Input parameters: geographical coordinates of each site, i.e. longitude λ, latitude φ, and altitude h:

[0048] Takeoff coordinates a t (λ t ,φ t , h t ), landing coordinate a l (λ l ,φ l , h l ), the coordinates of the mission target point Charging station coordinates

[0049] (2) Determine the matrix dimension and index:

[0050] Total number of sites N = 2 + n + k;

[0051] Index mapping: assign a unique index to each station, for example: 0: take-off point; 1: landing point; 2 to n+1: mission target point; n+2 to n+1+k: charging station;

[0052] (3) Calculate the flight distance between two stations

[0053] Calculate 3D Euclidean distance

[0054]

[0055] Since the flight distance is independent of the direction, the matrix satisfies d i,j =d j,i , that is, the matrix is ​​a symmetric matrix;

[0056] (4) Constructing distance matrix MD

[0057] Matrix structure: Generate an N×N symmetric matrix, where each element MD[i][j] represents the flight distance from station i to station j.

[0058] S2, based on the simulated annealing algorithm, generates an initial flight path, generates a new state by randomly exchanging the target position in the path, gradually reduces the temperature until convergence, and obtains a single calculation optimization solution, such as Figure 2 As shown, specifically including:

[0059] Initialization parameters of simulated annealing algorithm: set initial temperature T 初始 =100, crystallization temperature T 结晶 =0; temperature attenuation coefficient α = 0.01; define the number of iterations under isothermal state C = 30 × i, i is the number of target points, and the state stability counter threshold (JS = 30), the judgment logic of the state stability counter threshold (JS = 30) is: if the optimal total flight distance E is not updated after 30 consecutive iterative optimizations b , the iterative optimization at the current temperature is terminated early.

[0060] Generate initial route:

[0061] Fix the take-off and landing points, and randomly arrange the mission target points to form the initial flight route L O ;

[0062] Calculate the total flight distance E of the initial path O , and set the current route L c =L O 、Current energy E c =E O , and record the current optimal route L b =L c and optimal energy E b =E c .

[0063] Iterative optimization under isothermal conditions:

[0064] Randomly exchange nodes: Generate two different random numbers X and Y (both between 1 and the number of target points), and exchange the current route Lc The task points at the X and Y positions in the game are used to generate a new route.

[0065] Calculate the total flight distance E' of the new route c , and calculate the energy difference ΔE=E' c -E b .

[0066] Metropolis Criteria Judgment:

[0067] If ΔE<0 (the new path is better), directly accept the new route and update L b =L' c and E b =E' c ;

[0068] If ΔE ≥ 0 (the new path is worse), accept the route with probability P(ΔE) = exp(-ΔE / (kT)) (k is a constant, T is the current temperature).

[0069] State stability judgment: Repeat the above exchange and judgment process until the number of isothermal iterations C at the current temperature is completed, or no better solution is found after JS consecutive iterations, then the state is considered stable.

[0070] Temperature decay and cycling:

[0071] Reduce the temperature according to the attenuation coefficient α: T 新 =T 当前 ×(1-α)

[0072] If the current temperature T 当前 >T 结晶 , return to step 4 for a new round of iteration; otherwise, terminate the algorithm.

[0073] Output:

[0074] Finally output the current optimal route L b And its corresponding total flight distance E b .

[0075] S3. Run the simulated annealing algorithm at least three times independently to generate multiple single-calculation optimization solutions. If any two solutions are consistent and are the current optimal solutions, they are determined as the global optimal solution. If they are inconsistent, only the current optimal solution is retained and the simulated annealing algorithm is rerun until any two solutions are consistent and are the optimal solutions.

[0076] In the specific implementation, let R1, R2, and R3 be the initial results generated by the three simulated annealing algorithms. Assuming R1 is the best among the three, if R1 = R2 or R1 = R3, then R1 is the final output. If the above conditions do not hold, then R2 and R3 are eliminated, and the simulated annealing algorithm is called again to generate R2 and R3, and the comparison is repeated. It is important to note that the subscripts in R1, R2, and R3 do not represent the order of the data; they are only used to mark the initial results generated by the three simulated annealing algorithms.

[0077] S4. Output the optimal solution: The output results include the sequence of stations arranged in flight order, the flight distance of each section and the total mileage, and the charging station insertion location is marked. Among them, the take-off point of the optimal solution flight route is the route starting point, the landing point is the route end point, each mission target is visited only once, and the flight distance between any adjacent stations does not exceed the current remaining range of the drone.

[0078] In the above, the state stability judgment process of the simulated annealing algorithm is:

[0079] Repeat the iterative optimization and Metropolis criterion judgment until the number of isothermal iterations C at the current temperature is completed, or no better solution is found after JS consecutive iterations.

[0080] Note that the takeoff and landing points are fixed; only the order of intermediate mission objectives and charging stations is adjustable. The round-trip distance from the current location to the nearest charging station plus the planned flight distance must be less than or equal to the maximum flight distance on a single charge. Charging stations are allowed to be inserted into the flight route, and re-accessible charging stations are permitted. Insertion locations must ensure that the flight distance between adjacent stations does not exceed the drone's endurance.

[0081] While various embodiments of the present invention have been described above, the above descriptions are intended to be illustrative, non-exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or technological improvements in the marketplace, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for planning the shortest flight path for a multi-objective UAV, characterized in that: The following steps are involved: Obtain the coordinates of the drone's maximum range, takeoff point, landing point, mission target point, and charging station, and generate a distance matrix containing the distances between all stations; The initial flight path is generated based on the simulated annealing algorithm. New states are generated by randomly exchanging target positions in the path. The temperature is gradually reduced until convergence, resulting in a single-calculation optimization solution. Run the simulated annealing algorithm at least three times independently to generate multiple single-calculation optimization solutions. If any two solutions are consistent and are the current optimal solutions, they are determined to be the global optimal solution. If they are inconsistent, only the current optimal solution is retained and the simulated annealing algorithm is rerun until any two solutions are consistent and are the optimal solutions. Output the optimal solution: including flight path and total mileage information.

2. The method for planning the shortest flight path of a multi-objective UAV according to claim 1, characterized in that: The take-off point of the optimal solution flight route is the route starting point, and the landing point is the route ending point; Each mission target is visited only once; and the flight distance between any adjacent stations does not exceed the current remaining range of the drone.

3. The method for planning the shortest flight path of a multi-objective UAV according to claim 2, characterized in that: The initial flight route generation process is as follows: fix the positions of the take-off point and the landing point, randomly arrange the mission target points, and form the initial flight route L O ; Calculate the total flight distance E of the initial flight route O , and set the current flight route L c =L O 、Current total flight distance E c =E O , and record the current optimal route L b =L c and the optimal total flight distance E b =E c .

4. The method for planning the shortest flight path of a multi-objective UAV according to claim 3, characterized in that: In the simulated annealing algorithm, the initial temperature T 初始 =100, crystallization temperature T 结晶 =0; Temperature attenuation coefficient α = 0.01; Define the number of iterative optimizations under isothermal state C = 30 × i, the state stability counter threshold JS = 30, where i is the number of mission target points, and the judgment logic of the state stability counter threshold (JS = 30) is: if the optimal total flight distance E is not updated after 30 consecutive iterative optimizations b , the iterative optimization at the current temperature is terminated early.

5. The method for planning the shortest flight path of a multi-objective UAV according to claim 4, characterized in that: In the simulated annealing algorithm, the temperature decay process is: Reduce the temperature according to the attenuation coefficient α: T 新 =T 当前 ×(1-α) If the current temperature T 当前 >T 结晶 , conduct a new round of iterative optimization; Otherwise, terminate the algorithm; Among them, T 当前 Indicates the temperature value that the algorithm is using in a certain round of iteration, T 新 Represents the updated temperature value after temperature decay, that is, the temperature to be used in the next iteration.

6. The method for planning the shortest flight path of a multi-objective UAV according to claim 5, characterized in that: The iterative optimization process of the simulated annealing algorithm under isothermal conditions is: Random exchange node: Generate two different random numbers X and Y, and exchange the current flight path L c The task points at the X and Y positions in the , generate a new route L' c ; Calculate the total flight distance E' of the new route c , and calculate the distance difference ΔE=E' c -E b ; Accept or reject the new path L' according to the Metropolis criterion c .

7. The method for planning the shortest flight path of a multi-objective UAV according to claim 6, characterized in that: The Metropolis criteria are: If ΔE<0, directly accept the new route and update L b =L' c and E b =E' c ; If ΔE ≥ 0 (the new path is worse), accept the new path L' with probability P(ΔE) = exp(-ΔE / (kT)) c , where k is a constant and T is the current temperature.

8. The method for planning the shortest flight path of a multi-objective UAV according to claim 7, characterized in that: The state stability judgment process of the simulated annealing algorithm is: Repeat the iterative optimization and Metropolis criterion judgment until the number of isothermal iterations C at the current temperature is completed, or no better solution is found after JS consecutive iterations.

Images