Unmanned aerial vehicle route planning method based on simulated annealing-crown porcupine optimization algorithm

Abstract

The invention discloses an unmanned aerial vehicle route planning method based on a simulated annealing-crown porcupine optimization algorithm, is applied to the technical field of unmanned aerial vehicle task planning, and aims to solve the problems of low efficiency and low planning accuracy when solving a route planning problem in the prior art. The method comprises the following steps: firstly, based on the actual demand and distance of a flight mission, establishing an unmanned aerial vehicle route planning model containing a range constraint; an initial parameter set of an improved crown porcupine algorithm is defined, a population is initialized through a Latin hypercube sampling strategy, and the diversity and coverage range of the population are ensured. Then, simulating the foraging behavior of the crown porcupines to construct an optimization model, and simulating the route search of the unmanned aerial vehicle in a variable environment; and finally, an optimal route planning sequence is obtained through iterative solution, and efficient task planning of the unmanned aerial vehicle under the limitation of the voyage is realized. According to the method, the global search capability and the solving speed of route planning are remarkably improved, and the method is suitable for the optimal route planning problem of the unmanned aerial vehicle in a complex environment.

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned aerial vehicle mission planning, and in particular relates to a unmanned aerial vehicle route planning technology based on a simulated annealing-crown porcupine optimization algorithm. Background Art

[0002] Unmanned Aerial Vehicles (UAVs) play an increasingly critical role in contemporary technological innovation. Their practical applications in multiple fields, including but not limited to aerial photography, border surveillance, environmental assessment, material distribution, and disaster response, have demonstrated the versatility and practicality of drone technology. With the continuous advancement of technology, the performance boundaries of drones have been greatly expanded. They are now able to carry more advanced sensors, have longer flight times and higher flight stability, and can perform missions in adverse weather conditions.

[0003] In the practical application of UAVs, path planning is a key link to ensure efficient execution of tasks. Path planning not only involves the design of the flight trajectory of the UAV from the starting point to the end point, but also includes careful consideration of various environmental factors and mission requirements, such as avoiding obstacles, adapting to terrain changes, considering flight safety and energy efficiency, etc. However, traditional path planning methods mostly rely on pre-set flight plans and lack the ability to adapt to real-time environmental changes, which limits the autonomous decision-making and response speed of UAVs in dynamic environments.

[0004] Existing UAV path planning algorithms can be divided into three main categories: graph search methods, sampling-based methods, and bio-inspired methods. Graph search methods, such as A* and Dijkstra algorithms, are valued for their ability to find optimal paths in theory, but they face problems of high computational complexity and high memory consumption when dealing with large-scale data sets. Sampling-based methods, such as rapid random trees (RRT) and their variants, can handle high-dimensional space problems, but they usually lack consideration of path smoothness and optimization, and may not guarantee to find a feasible path in complex environments. Bio-inspired methods, which imitate the behavior patterns of organisms in nature, such as genetic algorithms, particle swarm optimization, and ant colony algorithms, are favored in path planning due to their strong adaptability and parallel processing capabilities, especially when dealing with dynamic environments and multi-objective optimization problems. However, it is worth noting that path planning algorithms inspired by biological intelligence still have the following shortcomings:

[0005] (1) The algorithm is prone to being trapped in a local optimal solution. Although biologically inspired methods in UAV path planning, such as genetic algorithms, can effectively explore a broad solution space, they may be trapped in a local optimal solution when faced with complex terrain and dynamic obstacles. Since the algorithm relies on random mutations during the iteration process, this randomness is sometimes not enough to guide the algorithm out of the local optimal area, making it impossible to guarantee the finding of the global optimal or suboptimal path.

[0006] (2) Algorithm parameter adjustment is more complicated. Bio-inspired algorithms require a large number of parameters to be adjusted for specific problems, such as the number of particles, inertia weight, and speed limit in particle swarm optimization, and the pheromone volatilization rate and update rules in ant colony algorithms. The sensitivity of these parameters requires algorithm designers to have a deep understanding and rich experience to ensure that the algorithm can run stably and achieve good performance in the UAV path planning problem. Improper parameter settings may cause the algorithm to converge prematurely or fail to converge, affecting the quality and efficiency of path planning.

[0007] (3) The algorithm has high requirements for computing resources. Bio-inspired methods may consume a lot of computing resources when dealing with UAV path planning problems, especially in large-scale or high-dimensional spaces. The algorithm's iterative process requires the evaluation of a large number of potential solutions, which not only increases the computing time but also places higher demands on memory and processors. In real-time path planning or resource-constrained environments, this high resource consumption may become a bottleneck restricting the application of the algorithm, limiting its widespread deployment in practical applications. Summary of the invention

[0008] In order to solve the above technical problems, the present invention proposes a UAV route planning method based on simulated annealing-crown porcupine optimization algorithm, which can effectively improve the global search capability and solution speed of route planning.

[0009] The technical solution adopted by the present invention is: a UAV route planning method based on simulated annealing-crown porcupine optimization algorithm, comprising:

[0010] S1. Model the UAV route planning scenario based on graph theory knowledge, establish the objective function and its constraints, and build a mathematical model for route planning;

[0011] S2, setting the initialization parameters of the crested porcupine algorithm, including population size, number of iterations, number of cyclic population cycles, etc.;

[0012] S3, looping and iterating the optimization process; setting problem-solving parameters and iteration end conditions; recording the global optimal solutions in the solution space, i.e., the optimal positions; simulating four different defense behaviors of crested porcupines facing predators in the natural environment and establishing an optimization iteration method in combination with a hybrid optimization strategy; after one cycle, updating the local optimal solution, i.e., the individual position, according to the simulated annealing method; until the iteration end conditions are met, jumping to step S4;

[0013] S4. Solve to obtain the optimal route planning scheme for the UAV and obtain the optimal path of the UAV based on the waypoints.

[0014] Beneficial effects of the present invention: First, a mathematical model of the route planning problem is established, and then a simulated annealing-crested porcupine optimization algorithm is designed to solve it. The improved Crested Porcupine Optimizer (CPO) algorithm significantly enhances population diversity and global search capabilities by introducing Latin hypercube sampling to generate uniformly distributed initial solutions, dynamically adjusts exploration and development weights, and performs a simulated annealing local search mechanism after each update of the global best solution. At the same time, it improves local development capabilities and overall convergence speed, making the algorithm more efficient and accurate in UAV path planning in complex environments;

[0015] The original crested porcupine algorithm has some limitations, such as uneven distribution of initial solutions, inflexible adjustment of exploration and development weights, and easy to fall into local optimal solutions. These factors may lead to limited efficiency and accuracy of the algorithm in UAV path planning in complex environments. The present invention significantly improves the performance of the UAV path planning algorithm through the following innovative strategies: first, the Latin hypercube sampling method is used to generate the initial solution to ensure the uniform distribution of the solution; second, a dynamic weight adjustment mechanism is introduced to balance the proportion of exploration and development; third, each time the global optimal solution is found, the simulated annealing algorithm is started for local search to refine the quality of the solution. These measures work together to not only enhance the local search capability of the algorithm in complex environments and increase population diversity, but also accelerate the convergence speed, ensuring the efficiency and accuracy of path planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of the UAV route planning method of the improved crested porcupine optimization algorithm based on the hybrid strategy of the present invention.

[0017] Figure 2 Flowchart of the iterative portion of the present invention.

[0018] Figure 3 Flowchart of the local search process of the iterative part of the present invention.

[0019] Figure 4It is the optimal track curve diagram of the embodiment.

[0020] Figure 5 It is a graph of algorithm convergence of the embodiment. DETAILED DESCRIPTION

[0021] To facilitate those skilled in the art to understand the technical content of the present invention, the present invention is further explained below with reference to the accompanying drawings.

[0022] like Figure 1 As shown, the present invention is a UAV route planning method based on a hybrid strategy-improved crested porcupine optimization algorithm, comprising the following steps:

[0023] S1. Model the UAV route planning scenario based on graph theory knowledge, establish the objective function and its constraints, and build a mathematical model for route planning.

[0024] S11. Assume that the mission scenario graph G is a square area of ​​m×m. Set the starting point, the end point and the no-fly zone in G.

[0025] S12. Construct the cost function f of the drone path length cost L

[0026]

[0027] Among them, the drone track consists of n nodes; let the current path point be p i =(x i ,y i ), the node p i , node p i+1 The Euclidean distance between

[0028] S13. When the UAV enters the radar interference area, construct the cost function f of the radar threat cost R As shown:

[0029]

[0030]

[0031] Among them, d i is the distance from the current node to the center of the radar, d rmin is the minimum detection radius of the rth radar detection area, d rmax is the maximum detection radius of the rth radar detection area.

[0032] S14. When the distance between the drone and the obstacle center is less than the obstacle radius, a collision will occur. Therefore, it is necessary to keep a certain distance between the track node and the obstacle and construct the cost function f of the no-fly zone cost. C, as shown below:

[0033]

[0034]

[0035] Among them, p i is the current path point, Indicates that from p i to p i+1 The vector, C i ,i=1,…,n are obstacles, represents the empty set, and inf represents infinity;

[0036] S15. When a drone performs a task, its energy consumption depends on power and flight time. Since the power of a drone is not constant during the entire flight process, let the uniform flight speed of the drone be v1(t), the straight flight distance be L1, the power be P1(t), in m / s; the turning curve flight distance be L2, the power be P2(t), then the total energy consumption cost of the drone during the flight is

[0037]

[0038] in, D is air resistance, in N; C D is the drag coefficient; ρ is the air density in kg / m 3 ; A is the windward area of ​​the drone, in kg / m 2 ; P2(t)=P1(t)+P extra , T extra =m·a c , m is the mass of the drone, in kg; a c is the centripetal acceleration of curved flight, r is the turning radius, in meters;

[0039] S16. Integrate and weigh the four factors of path length, radar threat, no-fly zone and energy consumption to model the fitness function. The cost function of the drone is as follows:

[0040] f=δ1f L +δ2f R +δ3f C +δ4f P

[0041] Among them, f is the total cost function of the track; f L is the cost function of the path length cost, f R is the cost function of radar threat cost, f Cis the cost function of the no-fly zone cost, f P is the energy price function; δ i ,i=1,2,3,4 are weight parameters. In the simulation example, δ1=0.5,δ2=0.1,δ3=0.2,δ4=0.2 are set, which means that the path length cost is mainly considered.

[0042] S2. Set the initialization parameters of the crested porcupine algorithm, including population size, number of iterations, number of cyclic population cycles, etc.:

[0043] S21. Define the initial population size N′ and the minimum population size N min , the number of cyclic population cycles T′, the convergence rate and the equilibrium coefficient T f . The population represents the candidate solutions of the path plan.

[0044] S22, Latin hypercube sampling initializes the population position.

[0045]

[0046] Where P i represents the position vector of the ith individual, L and U represent the lower and upper bounds of the solution space. i Represents the integer arrangement from 0 to N-1, ensuring that each interval has only one sample, r i Represents a random number vector in the interval [0,1], which is used to increase the randomness of sampling. N′ represents the population size.

[0047] The initial population can be expressed in the following matrix form:

[0048]

[0049] Where n is the dimension of the route planning problem, that is, the number of waypoints.

[0050] S3, iterate the optimization process, such as Figure 2 As shown, it includes the following steps:

[0051] Set the maximum number of iterations, record the global optimal solution in the solution space, that is, the optimal position; simulate four different defense behaviors of crested porcupines facing predators in the natural environment and establish an optimization iteration method in combination with a hybrid optimization strategy. After one cycle, update the local optimal solution, that is, the individual position, according to the simulated annealing method. When the iteration end condition is met, jump to S4.

[0052] S31, calculate the current optimal position in the environment solution space According to the current number of iterations t and the current population size N′, the population size is updated based on the cyclic population reduction strategy. The update method is as follows:

[0053]

[0054] S32, traverse each crested porcupine individual in the population, and each individual performs exploration (visual defense and sound defense) or exploitation (odor attack and physical attack) according to the adaptive weight strategy. Before the exploration or exploitation phase, the following parameters of each individual need to be updated: odor diffusion rate

[0055]

[0056] Where f(P i t ) is the objective function value of the ith individual at time t, ∈1 is set to 2.2204×10 in the simulation example to avoid a small value of 0 in the denominator. -16 .

[0057] Predator quality

[0058]

[0059] Where ∈2 is set to 2.2204×10 to avoid small values ​​of the denominator being 0. -16 .

[0060] The average force F of the predator on the crested porcupine i t :

[0061]

[0062]

[0063]

[0064] where τ is a random vector between 0 and 1, is the initial velocity of the ith individual at time t, is the final velocity of a randomly selected individual j at iteration t, a random solution is selected from the current population, and Δt is the number of current iterations.

[0065] The fitness change rate and population diversity are calculated in each iteration, and the weights of exploration and exploitation are dynamically adjusted based on these indicators.

[0066]

[0067]

[0068]

[0069]

[0070] W exploit =1-W explore

[0071] GB.Cost is the global optimal cost, PB.Cost is the optimal cost of the previous generation, and ∈ is set to 2.2204×10 in the simulation example to avoid small values ​​of 0 in the denominator. -16 ; Mean is the average value of all costs of the current population, W explore is the exploration weight, W exploit To develop weights.

[0072] Then, the exploration or development phase is decided according to the adaptive weight strategy: a random number τ1 is generated, when τ1 <W explore When τ1≥W explore Enter S35.

[0073] S33, Defense strategy 1: visual defense. Generate two uniformly distributed random numbers, τ3, τ4. When τ3<τ4, execute the first defense strategy, visual defense. When the crested porcupine detects a predator, it will raise its spikes as a warning. The predator has two choices: approach the crested porcupine or turn to an unexplored area. To simulate this behavior, the position update method is as follows:

[0074]

[0075] Among them, τ5 is a random number based on normal distribution, τ6 is a random value in the interval [0,1], is the vector generated between the current crested porcupine and a randomly selected crested porcupine from the population, used to represent the position of the predator at time t,

[0076]

[0077] When τ3≥τ4, the second defense strategy, sound defense, is executed, and enters S34.

[0078] S34. Defense strategy 2: Sound defense. When the crested porcupine makes a warning sound to deter a predator, the volume increases as the predator approaches. The predator's response to this threat can be divided into three types: one is to approach the crested porcupine; the second is to remain still; the third is to move away from the crested porcupine. When simulating this behavior, the position update rule is as follows:

[0079]

[0080] Among them, τ7 is a random value in the interval [0,1], r1 and r2 are random integers between [1,N], and U1 is a randomly generated binary vector of 0, 1.

[0081] S35, Defense strategy three: Smell attack. Generate a random number τ8 uniformly distributed between (0,1). When τ8 <T f When , the third defense strategy, odor attack, is executed. In this strategy, the crested porcupine will secrete a foul odor that spreads to the area around it to prevent predators from approaching it. To simulate this behavior, the position update method is as follows:

[0082]

[0083] Among them, r1 is a random integer between [1, N], δ is a parameter vector used to control the search direction, and γ t It is a defense factor.

[0084]

[0085] Here, τ9 is a random vector uniformly distributed between (0,1).

[0086]

[0087] Among them, τ 10 is a random value in the interval [0,1].

[0088] When U1 is equal to 0, the predator stops moving due to fear, and the crested porcupine also stops releasing odor, and the distance between the two remains unchanged. When U1 is 1, the crested porcupine will release a large amount of odor due to the approach of the predator. When U1 is between 0 and 1, the predator and the crested porcupine keep a safe distance, and the crested porcupine does not need to release a large amount of odor.

[0089] When τ8 ≥ T f When the fourth defense strategy, sound defense, is executed, the process goes to S36.

[0090] S36, Defense Strategy 4: Physical attack. When a predator gets too close, the crested porcupine will use its quills to collide with the predator. During this collision, the two are in close contact, which can be regarded as a physical phenomenon of one-dimensional inelastic collision. To simulate this behavior, the position update method is as follows:

[0091]

[0092] where τ 11 ,τ 12 is a random value in the interval [0,1].

[0093] S37, local search mechanism. After each update of the global best solution, the simulated annealing algorithm (SA) is used to perform local search to further optimize the current global best solution. Figure 3 As shown in the figure, the simulated annealing algorithm starts from the initial solution and iterates to generate neighboring solutions, i.e. new solutions. If the new solution has a lower cost than the current solution or meets certain probability conditions (i.e. the exponential ratio of the cost difference between the new and old solutions and the temperature is less than the random number), it will be adopted; otherwise, the temperature of the next iteration will be updated according to the current temperature T*α. With the iteration, the temperature exponentially decays, reducing the probability of accepting inferior solutions, thereby gradually approaching the optimal solution.

[0094]

[0095] Among them, P is the probability of accepting a new solution, T is the current temperature, and P newCost and P currentCost are the costs of the new solution and the current solution, P newCost and P currentCost Specifically, it is calculated according to the drone cost function formula in step S16. This process is repeated until a predetermined number of iterations is reached or the temperature drops sufficiently low, and finally a solution improved by local search is output.

[0096] S4. Based on the optimal solution improved by local search in S37, a set of waypoint coordinates from the starting point to the end point that satisfies the route cost function is obtained, that is, the optimal route planning scheme for the UAV.

[0097] In this embodiment, the task scene is a rectangular area G, which contains three circular dangerous areas and three rectangular dangerous areas D = {D i}, i = 1, 2, 3, 4, 5, 6, the coordinates of the center of the circular danger zone are (6.5, 1.0), (4.0, 3.0), (1.2, 1.5), the radius is 1.5, 1.0, 0.8, the starting point is (0, 1), and the end point is (8, 4.5). The coordinates of the rectangular danger zone are (6.0, 5, 1.5, 2), (1.5, 4, 2, 2.0), (3.5, 0.5, 1.5, 1.5).

[0098] In this example, the total number of iterations of the optimization model is T = 500, the initial population size is N′ = 200, and the minimum population size is N min = 150, the number of cyclic population periods T′ = 2, the convergence rate α = 0.2 and the equilibrium coefficient T f =0.75.

[0099] According to the above mission scenarios and parameter settings, the optimal route trajectory is calculated as follows: Figure 4 As shown, the algorithm converges as follows Figure 5As shown in the figure, it can be seen that the algorithm converges to the global optimum. The simulation results confirm that the UAV route planning method based on the hybrid strategy improved crown porcupine optimization algorithm can realize the generation of the optimal route for the UAV.

[0100] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. For those skilled in the art, the present invention may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A UAV route planning method based on simulated annealing-crown porcupine optimization algorithm, characterized in that: include: S1. Model the UAV route planning scenario based on graph theory knowledge, establish the objective function and its constraints, and build a mathematical model for route planning; S2, setting the initialization parameters of the crested porcupine algorithm, including at least the population size, the number of iterations, and the number of cyclic population cycles; S3, looping and iterating the optimization process; setting problem-solving parameters and iteration end conditions; recording the global optimal solutions in the solution space, i.e., the optimal positions; simulating four different defense behaviors of crested porcupines facing predators in the natural environment and establishing an optimization iteration method in combination with a hybrid optimization strategy; after one cycle, updating the local optimal solution, i.e., the individual position, according to the simulated annealing method; until the iteration end conditions are met, jumping to step S4; S4. Solve to obtain the optimal route planning scheme for the UAV and obtain the optimal path of the UAV based on the waypoints.

2. A UAV route planning method based on simulated annealing-crown porcupine optimization algorithm according to claim 1, characterized in that: Step S1 specifically includes the following sub-steps: S11, let the mission scenario graph be G, and set the starting point, end point and no-fly zone in G; S12. Construct the cost function f of the drone path length cost L : Among them, the UAV track consists of n nodes. Represents node p i With node p i+1 The Euclidean distance between them, i=1,…,n; S13. When the UAV enters the radar interference area, construct the cost function f of the radar threat cost R As shown: Among them, d i is the distance from the current node to the center of the radar, d rmin is the minimum detection radius of the rth radar detection area, d rmax is the maximum detection radius of the rth radar detection area; S14. When the distance between the drone and the obstacle center is less than the obstacle radius, a collision will occur. Therefore, it is necessary to keep a certain distance between the track node and the obstacle and construct the cost function f of the no-fly zone cost. C , as shown below: where p i is the current path point, C i For obstacles, represents the empty set, and inf represents infinity; S15. Assume that the flight speed of the drone is v1, the straight flight distance is L1, the power is P1(t), the turning curve flight distance is L2, and the power is P2(t). During the flight, the total energy consumption cost of the drone is: S16. Integrating and weighing the four factors of path length, radar threat, no-fly zone, and energy consumption, the cost function of the drone is as follows: f=δ1f L +δ2f R +δ3f C +δ4f P Among them, f is the total cost function of the track, f L is the cost function of the path length cost, f R is the cost function of radar threat cost, f C is the cost function of the no-fly zone cost, f P is the energy price function.

3. A UAV route planning method based on simulated annealing-crown porcupine optimization algorithm according to claim 2, characterized in that: Step S3 specifically The following steps are included: S31. According to the current number of iterations t and the current population size N′, the population size is updated based on the cyclic population reduction strategy. The updating method is as follows: S32. Traverse each crested porcupine individual in the population, and each individual performs exploration or development according to the adaptive weight strategy; before the exploration or development phase, the following parameters of each individual need to be updated: Odor diffusion rate Predator quality Average force of predators on crested porcupines Among them, P i t represents the position vector of the ith individual at time t, f(P i t ) is the objective function value of the ith individual at time t, ∈1 is to avoid small values ​​with a denominator of 0; ∈2 is to avoid small values ​​with a denominator of 0; τ is a random vector from 0 to 1, is the initial velocity of the ith individual at time t, is the final velocity of a randomly selected individual j at iteration time t, a random solution assignment is selected from the current population, and Δt is the number of current iterations; S33, Defense strategy 1: visual defense; Generate two uniformly distributed random numbers, τ3, τ4, when τ3 < τ4, execute the first defense strategy, visual defense; When the crested porcupine detects a predator, it will raise its spikes as a warning, and the predator has two choices: approach the crested porcupine, or turn to an unexplored area; Simulate this behavior, the position update method is as follows: Among them, τ5 is a random number based on normal distribution, τ6 is a random value in the interval [0,1], is the vector generated between the current crested porcupine and a randomly selected crested porcupine from the population, used to represent the position of the predator at time t; When τ3≥τ4, the second defense strategy, sound defense, is executed, and S34 is entered; S34. Defense strategy 2: sound defense. When the crested porcupine makes a warning sound to deter a predator, the volume increases as the predator approaches. The predator's response to this threat can be divided into three types: one is to approach the crested porcupine; the second is to remain still; the third is to move away from the crested porcupine. When simulating this behavior, the position update method is as follows: Among them, τ7 is a random value in the interval [0,1], r1 and r2 are random integers between [1,N], and U1 is a randomly generated binary vector of 0, 1; S35, Defense strategy three: Smell attack; Generate a random number τ8 uniformly distributed between (0,1), when τ8 <T f When the porcupine is about 100 km away from the predator, it executes the third defense strategy, odor attack; in this strategy, the crested porcupine will secrete a foul odor that spreads to the area around it to prevent predators from approaching it; simulating this behavior, the position update method is as follows: Among them, r1 is a random integer between [1, N], δ is a parameter vector used to control the search direction, and γ t It is a defense factor; When τ8 ≥ T f When , the fourth defense strategy, physical attack, is executed, and the process goes to S36; S36, Defense Strategy 4: Physical attack; When the predator is too close, the crested porcupine will use its quills to collide with the body; During this collision process, the two are in close contact, which can be regarded as a physical phenomenon of one-dimensional inelastic collision; To simulate this behavior, the position update method is as follows: where τ 11 ,τ 12 is a random value in the interval [0,1]; S37, local search mechanism; after each update of the global best solution, use the simulated annealing algorithm to perform local search to further optimize the current global best solution; the simulated annealing algorithm starts from the initial solution and iteratively generates adjacent solutions, i.e., new solutions.

4. A UAV route planning method based on simulated annealing-crown porcupine optimization algorithm according to claim 3, characterized in that: The expression of δ is: Here, τ9 is a random vector uniformly distributed between (0,1).

5. The method for UAV route planning based on simulated annealing-crown porcupine optimization algorithm according to claim 3 is characterized in that: γ t The expression is: Among them, τ 10 is a random value in the interval [0,1].

6. The method for unmanned aerial vehicle route planning based on simulated annealing-crown porcupine optimization algorithm according to claim 3 is characterized in that: In the simulated annealing algorithm, a new solution is adopted if it has a lower cost than the current solution or meets certain probability conditions; the certain probability condition specifically refers to the exponential ratio of the cost difference between the new and old solutions and the temperature being less than the random number.

7. The method for UAV route planning based on simulated annealing-crown porcupine optimization algorithm according to claim 6 is characterized in that: The probability formula for adopting the new solution is: Among them, P is the probability of accepting a new solution, T is the current temperature, and P newCost and P currentCost are the costs of the new solution and the current solution respectively.

Images