Improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning

Abstract

The invention relates to the technical field of unmanned aerial vehicle logistics and path planning, in particular to an improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning. The method comprises the following steps: firstly, constructing a three-dimensional simulation environment containing an obstacle and setting a path constraint condition; a comprehensive objective function is established, and four sub-objectives of path length, collision penalty, smooth penalty and height violation penalty are integrated through weighted summation; in the algorithm level, Kent chaotic mapping is adopted to generate an initial population to improve diversity, control parameters are changed into a nonlinear adaptive updating strategy to balance global exploration and local development, and a particle swarm optimization algorithm idea is fused to introduce an adaptive inertia coefficient and a dynamic weight coefficient to improve a grey wolf position updating rule; and finally, performing iterative optimization through an improved hybrid algorithm, and outputting an optimal path. The method can effectively improve the quality and efficiency of path planning, and enables the generated path to be economical, safe and smooth.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle logistics and path planning, and particularly relates to an improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning. BACKGROUND

[0002] As an important part of the intelligent logistics system, the path planning technology of unmanned aerial vehicle logistics distribution has attracted widespread attention in recent years; in practical applications, unmanned aerial vehicles need to fly safely and efficiently in complex three-dimensional urban environments, which requires path planning methods to effectively handle multi-objective optimization problems, taking into account path length, obstacle avoidance ability, flight height and smoothness and other constraints; as a swarm intelligence optimization algorithm, the grey wolf optimization algorithm has been explored and applied in the field of path planning due to its simple structure, fewer parameters and good optimization ability.

[0003] With the increasing complexity of application scenarios and the continuous improvement of delivery efficiency requirements, researchers continue to work on improving and integrating basic algorithms to further enhance their optimization performance and solution quality under multi-dimensional constraints; in this context, introducing new population initialization strategies, improving parameter adjustment mechanisms and integrating the advantages of other intelligent algorithms have become a natural development trend in this technical field to develop path planning methods that are more suitable for complex logistics distribution scenarios. SUMMARY

[0004] The present application aims to solve the problems in the background art and proposes an improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning.

[0005] The technical solution of the present application: an improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning, comprising the following specific implementation steps: S1, construct a three-dimensional simulation environment containing obstacles and set path constraint conditions; S2, establish a comprehensive objective function, which integrates the total path length sub-objective, collision penalty item sub-objective, smoothness penalty item sub-objective and height violation penalty item sub-objective through weighted summation; S3, use Kent chaotic mapping to generate an initial grey wolf population to improve population diversity; S4, change the update strategy of the control parameter a of the grey wolf optimization algorithm from linear decrease to nonlinear adaptive update; S5, improve the grey wolf position update rule by integrating the idea of particle swarm optimization algorithm, introduce an adaptive inertia coefficient and introduce dynamic weight coefficients for alpha wolves, beta wolves and delta wolves; S6, execute the iterative optimization process based on the improved grey wolf optimization algorithm, and output the optimal path; S7, calculating a path smoothness penalty term, which is calculated based on the cosine of the angle between two vectors formed by three consecutive points on the path.

[0006] Preferably, constructing the three-dimensional simulation environment and setting the path constraints specifically includes: Setting a two-dimensional grid map of 100m x 100m; Setting the upper limit of the height of the three-dimensional space to 50 meters; Setting the flight safety height of the unmanned aerial vehicle to five meters; Configuring two types of static obstacles, buildings and trees, in the environment and setting their safety areas; Setting the starting point coordinates and the end point coordinates of the path planning; Setting the no-fly zone that the path needs to avoid as the safety range of all obstacles, requiring the distance between the path points and any obstacle to be no less than the safety distance of five meters; Setting the number of intermediate path points to eight, and the optimization dimension of the entire path planning to twenty-four.

[0007] Preferably, the comprehensive objective function is specifically: The total path length sub-objective is the cumulative sum of the spatial straight-line distances between all adjacent path points: wherein, represents the total path length; N represents the number of total path points; (x i ,y i ,z i ) represents the three-dimensional space coordinates of the i-th path point; (x i+1 ,y i+1 ,z i+1 ) represents the three-dimensional space coordinates of the i+1-th path point; The collision penalty term adds a fixed value of penalty for each violation when the path point enters the safety range of the obstacle; The height violation penalty term adds a fixed value of penalty for each violation when the path point is below the set safety height; The smoothness penalty term is calculated based on the cosine of the angle between adjacent path points: wherein, β represents the smoothness weight; n is the total number of path points; θ i represents the angle between two vectors; The comprehensive objective function is: wherein, Cost represents the total cost function; F l represents the total path length ; P coll represents the collision penalty term; P height ​​​represents a high violation penalty term; a, b, g represent weight coefficients.

[0008] Preferably, the initial grey wolf population is generated by using the Kent chaotic mapping, and the Kent chaotic mapping is specifically as follows: The chaotic variable value is generated by using the Kent chaotic mapping formula. The Kent chaotic mapping formula is as follows: ; value i represents the chaotic variable value generated in the i-th iteration; value i+1 represents the chaotic variable value generated in the i+1-th iteration. The population is initialized by using the ergodic characteristic that the generated chaotic variable value is uniformly distributed in the interval [0, 1].

[0009] Preferably, the updating formula in step S4 is as follows: ; value max a is an intermediate variable; t represents the iteration number; t i is the maximum iteration number.

[0010] Preferably, the improved grey wolf position updating rule is specifically as follows: ; ; value i (t) is the position vector of the i-th grey wolf in the t-th generation; value i (t+1) is the updated position vector of the i-th grey wolf in the t+1-th generation; w is an adaptive inertia coefficient; c1 and c2 are coefficients of position updating; r1 and r2 are random numbers with values in the interval [0, 1]; value αbest , value βbest , and value δbest are the position vectors of the optimal solution a wolf, the suboptimal solution b wolf, and the third optimal solution d wolf in the current solution space; |D α |, |D β |, and |D δ | are absolute values of the differences between the a wolf, the b wolf, and the d wolf and related objects.

[0011] Preferably, the adaptive inertia coefficient is dynamically adjusted according to the individual fitness, and the calculation formula is as follows: ; value is the adaptive inertia coefficient of the i-th grey wolf in the d-th iteration; , are the minimum and maximum inertia coefficients, respectively. fitness value of the i-th grey wolf at the d-th iteration; minimum fitness of all grey wolves at the d-th iteration; average fitness of all grey wolves at the d-th iteration.

[0012] Preferably, the iterative optimization process specifically comprises the following steps: a parameter initialization step of setting population size, maximum iteration number, map environment, start point coordinates and end point coordinates; a population initialization step of generating the position of the grey wolf population based on Kent chaotic mapping; a fitness calculation step of decoding each grey wolf individual into a potential path and calculating a fitness value according to a comprehensive objective function; a leader wolf determination step of determining the alpha wolf, beta wolf and delta wolf in the current population according to the fitness value; a termination condition judgment step of checking whether the current iteration number reaches the preset maximum iteration number; a grey wolf position updating step of updating the position of all grey wolf individuals according to an improved position updating formula and performing boundary correction; an optimal solution output step of outputting the optimal path represented by the alpha wolf.

[0013] Preferably, the path smoothness penalty term is specifically calculated as: for three consecutive points on the path, a forward vector and a backward vector are defined: the forward vector: the backward vector: a cosine value of the included angle between the forward vector and the backward vector is calculated, and the calculation formula is: ; wherein, θ i is the included angle of the two vectors; (x i , y i , z i ) represents the three-dimensional space coordinates of the i-th track point; (x i+1 , y i+1 , z i+1 ) represents the three-dimensional space coordinates of the i+1-th track point; (x i-1 , y i-1 , z i-1 ) represents the three-dimensional space coordinates of the i-1-th track point; the smoothness penalty term is calculated based on the cosine value, and the algorithm is forced to reduce sharp turns by penalizing (1−cosθ i ).

[0014] ​​Compared with the prior art, the above technical scheme of the present application has the following beneficial technical effects: The present application designs an improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning, which initializes the population by introducing Kent chaotic mapping, effectively improves the diversity and uniformity of the initial population, lays a firmer foundation for global optimization, and helps the algorithm to jump out of the local optimum; the linear decreasing strategy of the control parameter is improved to a nonlinear adaptive update, so that the algorithm can perform more sufficient global exploration in the early stage of iteration, and more fine local development in the later stage, thereby better balancing the exploration and development capacity; the idea of particle swarm optimization algorithm is fused and an adaptive inertia coefficient and a dynamic weight coefficient are introduced, which significantly enhances the guidance and flexibility of the grey wolf individual position update, and improves the convergence speed and solution accuracy of the algorithm; the comprehensive objective function constructed by the present application considers multiple key performance indicators such as path length, collision risk, flight height compliance and path smoothness, and provides a comprehensive and reasonable quantitative standard for algorithm evaluation of path advantages and disadvantages through weighted summation, thereby ensuring that the final generated flight path meets the economic requirements while having high safety and good flyability; through the synergistic effect of multiple strategies, the present application improves the quality of path planning and the comprehensive performance of the algorithm as a whole, making it show stronger robustness and practicality when facing complex three-dimensional distribution environment. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 A method flowchart of an improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning is proposed in the present application. Figure 2 A CPS-GWO algorithm flowchart is proposed in the present application. DETAILED DESCRIPTION

[0016] In embodiment one, an improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning is proposed as shown in the figure, which includes the following specific implementation steps: Figure 1 as shown, including the following specific implementation steps: S1, a three-dimensional simulation environment is constructed, a 100m*100m plane grid and a 0-60m height range are set, the starting point and the ending point coordinates are configured, and two types of static obstacles and their safety areas, i.e. buildings and trees, are arranged, and the safe flight height, the number of path points and the optimization dimension are set, to provide a basic scene and constraint conditions for path planning, specifically: A 100m*100m two-dimensional grid map is set, and the coordinate range of X axis and Y axis is 0 to 100 meters; The upper limit of Z axis height in three-dimensional space is set to 50 meters, but the Z axis range is limited to 0-60 meters when finally visualized, to ensure that all paths and obstacles are clearly visible; The flight safety height of the UAV is set to 5 meters, i.e., the height of the path point should not be lower than this threshold; Two types of static obstacles are configured in the environment: Building: Its shape is a rectangular prism, and its position and size are determined by defining its upper left corner starting coordinates (x0, y0), the width in the X direction, the length in the Y direction, and the building height; For example, the parameters of one building can be defined as 20, 20, 10, 10, 15; Tree: Its shape is a pyramid, and to avoid existing buildings, small obstacle trees are generated at random positions with sizes of 1x1 or 2x2 meters and heights between 3-6 meters. Small obstacles are not generated within 5 meters of obstacles such as buildings, and a total of 30 are generated in this simulation; The starting point of the path planning is set to be located in the open area at the edge of the city with coordinates (5, 5, 10) meters; the end point is located in the diagonal area with coordinates (95, 95, 10) meters; The safety range of the forbidden flight area to be avoided in the path setting is the safety distance of 5 meters between the path point and any obstacle; In the path planning, the number of intermediate path points is set to 8, and since each path point contains X, Y, and Z coordinate values, the optimization dimension of the entire path planning is 24.

[0017] S2, establish a comprehensive objective function, integrate the total path length, collision penalty term, smoothing penalty term, and height violation penalty term through weighted summation, and set the corresponding weight coefficients to balance the economy, safety, and smoothness of the path, providing quantitative standards for algorithm evaluation of path quality, specifically: A three-dimensional space segment accumulation model is used, i.e., the spatial straight-line distance between consecutive path nodes is summed segment by segment to construct a total path length evaluation system: ; Where, represents the total path length, i.e., the cumulative sum of the spatial straight-line distances between all adjacent path points; N represents the total number of path points; (x i ,y i ,z i ) represents the three-dimensional space coordinates of the i-th path point; (x i+1 ,y i+1 ,z i+1 ) represents the three-dimensional space coordinates of the i+1-th path point, which forms an adjacent segment with the i-th path point and is used to calculate the spatial straight-line distance between them; The smoothing penalty term P smoothThe penalty term is used to constrain the path turning angle, improve the stability of the UAV flight, and is calculated based on the cosine value of the angle between adjacent path points, and the formula is: ; wherein β represents a smoothing weight; n is the total number of path points; θ i represents the angle between two vectors; By optimizing each sub-target, balancing the path length, collision penalty, flight safety height, path smoothness and other parameters, and assigning weights to each sub-target, the total target function is obtained by weighted summation: ; wherein Cost represents the total cost function; F l represents the total path length , reflecting the efficiency of the path, and the smaller the value, the shorter the path and the lower the flight energy consumption; P coll represents the collision penalty term, when the path point enters the obstacle safety range (less than the set safety distance of 5 meters), each violation adds a penalty value of 1000, forcing the algorithm to avoid obstacles and ensure flight safety; P height represents the height violation penalty term, when the path point is lower than the set safety height (such as the ground or the top of the obstacle with a safety distance of less than 5 meters), each violation adds a penalty value of 500, ensuring the rationality of the flight height; α, β, γ represent weight coefficients.

[0018] S3, the initial gray wolf population is generated by using Kent chaotic mapping formula, and the traversal characteristics of uniform distribution in (0, 1) are used to replace the traditional random initialization method, so as to improve the diversity and uniformity of the initial population and provide better search starting point for subsequent iteration, that is: The initial population is initialized by using the discrete chaotic system Kent chaotic mapping to increase the diversity of the population in the algorithm: ; wherein value i represents the chaotic variable value generated by the i-th iteration; value i+1 represents the chaotic variable value generated by the i+1-th iteration; It should be noted that the probability density function of Kent mapping is uniformly distributed in the range of (0, 1) and has good ergodicity; by using this characteristic to generate the initial population, high-density and low-repetition sampling can be realized, which significantly improves the diversity of the initial population and provides more abundant search starting points for subsequent iterations, thereby reducing the probability of premature convergence.

[0019] S4, change the update strategy of control parameters from linear decrease to nonlinear adaptive adjustment, enhance global exploration by slow decline in the early iteration, and strengthen local development by rapid trend to zero in the later period, to more accurately match the complex path search process, that is: ; wherein a is an intermediate variable; t represents the iteration number; t max is the maximum iteration number.

[0020] S5, fuse the particle swarm optimization (PSO) algorithm idea, improve the position update rule of grey wolf, introduce an adaptive inertia coefficient in the new rule, and dynamically adjust its value according to the individual fitness; at the same time, introduce dynamic weight coefficients λ1, λ2, λ3 for α, β, δ wolves, and the values are calculated based on the distance ratio of them and the current grey wolf, so as to quantify the different influence degrees of the three in the position update, which is specifically: This embodiment will improve the update strategy of the position of the grey wolf, and the improvement method is as follows: ; wherein λ1, λ2, λ3 are dynamic weight coefficients, used to quantify the influence degree of α, β, δ wolves on the position update of the bottom grey wolf (ω wolf); X i (t) is the position vector (solution representation) of the i-th grey wolf in the t-th generation; X i (t+1) is the updated position vector of the i-th grey wolf in the t+1-th generation; ω is the inertia coefficient, used to adjust the inertia of the position update of the grey wolf (influence the degree of the current position on the next position); c1, c2 are the coefficients of the position update (similar to the control coefficient in the original grey wolf optimization algorithm GWO, balance the "exploration" and "development" process); r1, r2 are random numbers with values between 0 and 1, introduce randomness in the position update, and enhance the diversity of the search; X αbest , X βbest , X δbest are the position vectors of the optimal solution (α wolf), the suboptimal solution (β wolf), and the third optimal solution (δ wolf) in the current solution space; ; introduce an adaptive inertia coefficient ω, whose value is dynamically adjusted according to the individual fitness, and the strategy is as follows: ; ; ; wherein |D α |, |D β |, |D δ | are the absolute values of the differences or distances between α wolf, β wolf, δ wolf and the relevant objects (such as the current position of the grey wolf); Let be the adaptive inertia coefficient of the i-th gray wolf in the d-th iteration (dynamically adjusted according to the relationship between its own fitness and the average fitness of the population). , These are the minimum and maximum inertia coefficients, respectively. Let be the fitness value of the i-th gray wolf in the d-th iteration; Let be the minimum fitness of all gray wolves at the d-th iteration; Let be the average fitness of all gray wolves at the d-th iteration.

[0021] S6. Execute the complete execution flow of the improved hybrid algorithm (CPS-GWO algorithm) based on the Gray Wolf Optimization (GWO) and Particle Swarm Optimization (PSO) algorithms, such as... Figure 2 As shown, specifically: A1. Begin; A2. Parameter initialization: Set parameters such as population size, maximum number of iterations, map environment, start and end coordinates; A3. Initialize the population: Based on the Kent chaotic mapping in step 3, initialize the position of the gray wolf population; A4. Calculate fitness: Decode each individual gray wolf in the population, convert its location information into a potential path, and then calculate the fitness value of the path (including path length, collision penalty, smoothness, etc.) according to the comprehensive objective function defined in step S2. A5. Determine the leader wolf: Based on the fitness value, determine the α wolf (optimal solution), β wolf (second best solution), and δ wolf (third best solution) in the current population. A6. Determine the termination condition: Check if the current iteration count has reached the preset maximum iteration count; if it has, proceed to step A8; if not, continue to the next step. A7. Update the gray wolf positions: Update the positions of all individual gray wolves according to the improved position update formula in step S5; then, correct the updated positions to ensure that they are within the feasible range of the solution space; after completing the update, return to step A4. A8. Output the optimal solution: Output the current position represented by the α wolf, which is the optimal path found by the algorithm; A9. End.

[0022] S7. The degree of path turning is quantified by calculating the cosine of the angle between two vectors (forward and backward vectors) formed by three consecutive points on the path. The closer this cosine value is to 1, the smoother the path. This calculated value is directly used to calculate the smoothness penalty term in step S2. By penalizing (1-cosθᵢ), the algorithm is forced to reduce sharp turns, thereby improving the smoothness and flyability of the final generated path. Specifically: For three consecutive points P on the path i-1 (x i-1 ,y i-1 ,z i-1 ), P i (x i ,y i ,z i ), P i+1 (x i+1 ,y i+1 ,z i+1 Define two vectors: Forward vector: ; Backward vector: ; The formula for calculating the cosine of the angle between these two vectors is: ; Where, θ i The angle between the two vectors; It should be noted that cosθ i The value range of θ is [-1, 1]. When its value is close to 1, the included angle θ i When the angle is close to 0°, the path is close to a straight line, and the smoothness is optimal; when the value is close to 0, the included angle is close to 90°, and the path has a clear turn; when the value is close to -1, the included angle is close to 180°, the path is almost reversed, and the smoothness is worst. The cosine value of this included angle is directly used in the smoothing penalty term P in step S2. smooth The calculation is performed by penalizing (1−cosθ). i This forces the algorithm to reduce sharp turns and improve path smoothness.

[0023] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. An improved grey wolf optimization method for unmanned aerial vehicle logistics delivery planning, characterized in that, The method comprises the following specific implementation steps: S1, constructing a three-dimensional simulation environment containing obstacles and setting path constraint conditions; S2, establishing a comprehensive objective function, which integrates a path total length sub-objective, a collision penalty term sub-objective, a smoothness penalty term sub-objective and a height violation penalty term sub-objective through weighted summation; S3, generating an initial grey wolf population by using Kent chaotic mapping to improve population diversity; S4, changing the updating strategy of the control parameter a of the grey wolf optimization algorithm from linear decrease to nonlinear adaptive updating; S5, improving the grey wolf position updating rule by fusing the idea of the particle swarm optimization algorithm, introducing an adaptive inertia coefficient and introducing dynamic weight coefficients for alpha wolves, beta wolves and delta wolves; S6, executing an iterative optimization process based on the improved grey wolf optimization algorithm to output an optimal path; S7, calculating a path smoothness penalty term based on the cosine value of the included angle of two vectors formed by three consecutive points on the path.

2. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 1, wherein, The construction of the three-dimensional simulation environment and the setting of the path constraint conditions specifically comprise: setting a two-dimensional grid map of 100 m*100 m; setting the upper limit of the height of the three-dimensional space to 50 meters; setting the flight safety height of the unmanned aerial vehicle to five meters; configuring two types of static obstacles, buildings and trees, in the environment and setting their safety areas; setting the starting point coordinates and the end point coordinates of the path planning; setting the no-fly zone that the path needs to avoid as the safety range of all obstacles, and requiring that the distance between the path points and any obstacle be not less than a safety distance of five meters; setting the number of intermediate path points to eight, and the optimization dimension of the entire path planning to twenty-four.

3. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 2, characterized in that, The establishment of the comprehensive objective function specifically comprises: The path total length sub-target is the accumulation of the spatial straight line distances between all adjacent track points: ; wherein, represents the total path length; N represents the number of total waypoints; (x i ,y i ,z i ) represents the three-dimensional spatial coordinates of the i-th waypoint; (x i+1 ,y i+1 ,z i+1 ) represents the three-dimensional spatial coordinates of the i+1-th waypoint; the collision penalty term adds a fixed value of penalty for each violation when a path point enters the safety range of an obstacle; the height violation penalty term adds a fixed value of penalty for each violation when a path point is lower than the set safety height; the smoothness penalty term is calculated based on the cosine value of the included angle of adjacent path points: ； where β represents a smoothing weight; n is the total number of path points; θ i represents the angle between two vectors; the comprehensive objective function is: ； wherein Cost represents a total cost function; F l represents a total path length ; P coll represents a collision penalty term; P height represents a high violation penalty term; a, b, g represent weight coefficients.

4. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 3, characterized in that, The generation of the initial grey wolf population by using Kent chaotic mapping specifically comprises: iteratively generating chaotic variable values by using a Kent chaotic mapping formula; the Kent chaotic mapping formula is: ； wherein value i represents the value of the chaotic variable generated at the i-th iteration; value i+1 represents the value of the chaotic variable generated at the i+1-th iteration; performing population initialization by using the traversal characteristics of the generated chaotic variable values uniformly distributed in the interval [0, 1].

5. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 4, characterized in that, The updating formula in step S4 is: ； where a is an intermediate variable; t represents the iteration number; t max is the maximum iteration number.

6. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 5, characterized in that, The improvement of the grey wolf position updating rule specifically adopts the following formula: ； ； wherein λ1, λ2, λ3 are dynamic weight coefficients; X i (t) is the position vector of the i-th gray wolf in the t-th generation; X i (t+1) is the updated position vector of the i-th gray wolf in the t+1-th generation; ω is an adaptive inertia coefficient; c1, c2 are coefficients of position update; r1, r2 are random numbers with values in [0, 1]; X αbest , X βbest , X δbest are the position vectors of the optimal solution α wolf, the suboptimal solution β wolf and the third optimal solution δ wolf in the current solution space; |D α |, |D β |, |D δ | are the absolute values of the differences between the α wolf, the β wolf, the δ wolf and the related objects.

7. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 6, characterized in that, The adaptive inertia coefficient is dynamically adjusted according to the individual fitness, and its calculation formula is: ； wherein, is the adaptive inertia coefficient of the i-th grey wolf at the d-th iteration; , are the minimum and maximum inertia coefficients, respectively; is the fitness value of the i-th grey wolf at the d-th iteration; is the minimum fitness of all grey wolves at the d-th iteration; is the average fitness of all grey wolves at the d-th iteration.

8. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 7, characterized in that, The execution of the iterative optimization process specifically comprises the following steps: a parameter initialization step of setting the population size, the maximum number of iterations, the map environment, the starting point coordinates and the end point coordinates; an initialization population step of generating the positions of the grey wolf population based on Kent chaotic mapping; a fitness calculation step of decoding each grey wolf individual into a potential path and calculating the fitness value according to the comprehensive objective function; a leader wolf determination step of determining the alpha wolf, the beta wolf and the delta wolf in the current population according to the fitness value; a termination condition judgment step of checking whether the current number of iterations reaches the preset maximum number of iterations; a grey wolf position updating step of updating the positions of all grey wolf individuals according to the improved position updating formula and performing boundary correction. Output the optimal solution step, output the optimal path represented by alpha wolf.

9. The improved grey wolf optimization method for unmanned aerial vehicle logistics distribution planning according to claim 8, wherein, The path smoothing penalty term is specifically calculated as: For three consecutive points on the path, define a forward vector and a backward vector: Forward vector: ; Backward vector: ; Calculate the cosine value of the included angle between the forward vector and the backward vector, and the calculation formula is: ； where θ i is the angle between two vectors; (x i , y i , z i ) represents the three-dimensional spatial coordinates of the i-th track point; (x i+1 , y i+1 , z i+1 ) represents the three-dimensional spatial coordinates of the i+1-th track point; (x i-1 , y i-1 , z i-1 ) represents the three-dimensional spatial coordinates of the i-1-th track point; The smoothness penalty term is based on the cosine of the angle value, forcing the algorithm to reduce sharp turns by penalizing (1−cosθ i ).

Images