Cluster collaborative route planning method based on particle swarm longicorn beetles

Abstract

The invention discloses a cluster collaborative route planning method based on particle swarm beetle antennae. The cluster collaborative route planning method is divided into two levels of particle swarm optimization population and beetle antennae search algorithm re-optimization. In the first level, the whole particle swarm is optimized by using the particle swarm, and global search of a solution space is completed by using a parallel search strategy implied by an algorithm; and in the second level, re-optimizing an optimization result searched in the previous level by adopting a beetle antennae algorithm so as to improve the algorithm precision. The global optimal solution searched by the original particle swarm algorithm is optimized again in a certain neighborhood range, if a better solution is found, the solution is used as a new global optimal solution, and if the better solution is not found, the previous global optimal solution is still adopted. According to the method, the defect that a traditional PSO method is prone to being caught in local is well avoided, and the optimal strike path is obtained for optimization of the unmanned aerial vehicle. The method also provides a valuable reference for a cluster collaborative track planning method in other complex environments.

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned aerial vehicles, and in particular relates to a cluster collaborative trajectory planning method based on particle swarm beetle whiskers. Background Art

[0002] Particle Swarm Optimization (PSO) was jointly proposed by American social psychologist James Kennedy and electrical engineer Russell Eberhart in 1995. It was originally derived from simulating the foraging behavior of birds. Each bird is abstracted as a massless and volumeless particle, and the process of finding the optimal solution to the problem is regarded as the process of birds finding food. Through cooperation and competition between individuals, the search for the optimal solution to complex problems is realized. Particle Swarm Optimization is widely used in industry, agriculture, medicine, military and other fields due to its simple structure, few setting parameters, strong global convergence and easy implementation.

[0003] The Beetle Antennae Search (BAS) algorithm is a highly efficient search optimization algorithm newly proposed in 2017, also known as the Beetle Algorithm. The principle of the Beetle Antennae Search algorithm originally comes from the foraging behavior of the longhorn beetle. It does not need to know the specific location information of the food, but forages according to the strength of the food smell. It has a pair of antennae on its head, which can sense the odor intensity of the food. If the odor intensity perceived by the left antennae is greater than the odor intensity perceived by the right antennae, the longhorn beetle flies to the left, otherwise it flies to the right. After several times, the longhorn beetle can find the location of the food. Scientists are inspired by this and believe that the smell of food can be regarded as a function, and the value of this function at each point in three-dimensional space is different. The two whiskers of the longhorn beetle can collect the odor values ​​of two points near itself, so that the optimal value of the function and the corresponding optimal position can be found by continuous searching.

[0004] The characteristics of the particle swarm algorithm are simple procedures, easy implementation, and strong global optimization ability. Although it has many advantages, the algorithm itself still has many problems. For example, the convergence speed of the algorithm is slow in the later stage, and the solution accuracy cannot meet the requirements. The particle swarm mainly relies on the cooperation and competition between groups to search for the best, so when a particle finds that its position is the current optimal position, other particles will quickly move to this position and finally gather at this point. This may cause a problem. If the position of this point is not the global optimal point, but the local optimal point, the particle will not be able to conduct a new search and fall into the local optimal point. This is the so-called "premature" phenomenon. At the same time, due to the influence of the inertia weight factor on the entire particle swarm to "fly" to the optimal solution based on past experience, there is a certain probability that the particle will miss the optimal solution and fly divergently to the space far away from the optimal solution, which will cause the algorithm to fail to converge. Summary of the invention

[0005] In order to overcome the shortcomings of the prior art, the present invention provides a cluster collaborative trajectory planning method based on particle swarm beetle whiskers, which is divided into two levels: particle swarm optimization population and beetle whisker search algorithm re-optimization. At the first level, a particle swarm is used to optimize the entire particle population, and the implicit parallel search strategy of the algorithm is used to complete the global search of the solution space; at the second level, the beetle whisker algorithm is used to re-optimize the optimization result of the previous level search to improve the algorithm accuracy. The global optimal solution searched by the original particle swarm algorithm is optimized again within a certain neighborhood range. If a better solution is found, the solution is used as the new global optimal solution. If a better solution is not found, the previous global optimal solution is still used. The method of the present invention well avoids the local defects of the traditional PSO method, and obtains the optimal strike path for the optimization of the unmanned aerial vehicle. This method also provides a valuable reference for cluster collaborative trajectory planning methods in other complex environments.

[0006] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0007] Step 1: Set the optimization objective function as:

[0008]

[0009] Assume that the target search space is D-dimensional, the population size is N-dimensional, and the position vector of the i-th particle is recorded as X i =(x i1 ,x i2 ,...,x iD ), the velocity vector is denoted by V i =(v i1 ,v i2 ,...,v iD ), the historical optimal value of the i-th particle is recorded as pbest i =(p i1 ,p i2 ,...,p iD ), the global optimal value of the entire population in the target space is denoted as g best =(g1,g2,...,g D ); the particle position and velocity update formula are as follows:

[0010]

[0011] Among them, c1 represents the particle's own learning ability, c2 represents the social learning ability, r1 and r2 are uniform random numbers in the range of [0,1], t represents the current iteration number, i=1,2,...,N, N is the number of particles in the population, V it denote the position and velocity of the t-th iteration respectively, V i t+1 They represent the position and velocity of the t+1th iteration respectively; the inertia weight w represents the ability of the particle to maintain its original velocity;

[0012] The right side of formula (2) consists of three terms: the first term represents particle inertia, indicating the tendency of particles to maintain their previous speed; the second term represents the individual cognitive part, reflecting the particle's learning of its own historical experience, indicating that the particle approaches its own historical optimal position; the third term represents the social cognitive part, reflecting the group historical experience of collaborative cooperation and information sharing among particles, indicating that the particle approaches the group's optimal position;

[0013] The inertia weight w is adjusted using a linear decreasing strategy during the iteration process, and its calculation formula is as follows:

[0014] w=w max -t((w max -w min ) / t max ) (4)

[0015] Among them, G represents the maximum number of iterations, w max represents the maximum inertia weight; w min represents the minimum inertia weight; t represents the current number of iterations;

[0016] Step 2: The beetle whisker algorithm is modeled as follows: X represents the centroid position of the beetle, X l Indicates the left whisker position, X r represents the position of the right whisker, and d0 represents the distance between the two whiskers;

[0017] The direction of the beetle's head is arbitrary during the optimization process, so it is represented by an arbitrary direction vector dir, and its normalized form is as follows:

[0018]

[0019] The formula for updating the optimal position of the longhorn beetle at time t is:

[0020]

[0021] Where f(·) represents the fitness function, X t , P t and P t-1 They represent the position of the longicorn at time t, the optimal position and the optimal position at time t-1 respectively;

[0022] The formula for updating the positions of the left and right whiskers of the longicorn at time t is:

[0023]

[0024] The fitness function value is calculated based on the left and right whiskers, and the fitness values ​​of the left and right whiskers are compared to determine the next movement direction of the longhorn beetle. The formula for updating the position of the longhorn beetle is:

[0025]

[0026] Where step is the search step length, and step = c·d0, c is a constant;

[0027] Step 3: Cluster collaborative trajectory planning method based on particle swarm beetle whiskers;

[0028] Step 3-1: Initialize the parameters c1, c2, c, w, and Step in the algorithm, and randomly initialize the position X of each particle in the generated population. i (t), calculate the fitness value corresponding to each particle according to the objective function;

[0029] Step 3-2: Let the optimal position of the individual at this time be P i (t) = X i (t), find the global optimal position P at this time g (t);

[0030] Step 3-3: Perform the following operations on each particle in the population: Update the velocity and position of the particle according to equations (2) and (3), and calculate the fitness value corresponding to the updated position; if the fitness value of the new position of the particle is better than the initial individual optimal value, update the individual optimal position and set P i (t) = X i (t); If the fitness value of the particle's new position is better than the global optimal value, then update the global optimal position, and let P g (t) = X i (t);

[0031] Step 3-4: Use the beetle whiskers to find the global optimal solution P for the particles g (t) Perform local search and set the initial solution of the beetle to P = P g (t) and the optimal solution P gbest =P g (t);

[0032] Step 3-5: Determine whether the termination condition of the beetle search is met. If the termination condition is met, terminate P g (t) is the beetle whisker search process, and the optimization result P is output at the same time gbest , update P gbest =P g (t) and jump to step 3-8; otherwise, jump to the next step;

[0033] Step 3-6: Randomly generate the direction vector dir of the beetle's whiskers, and update the positions P of the left and right whiskers according to equations (7) and (8) left and P right ;

[0034] Step 3-7: According to the updated position P left and P right Calculate the corresponding fitness value. Update the position of the next beetle individual according to formula (9);

[0035] Step 3-8: Determine whether the iteration limit has been reached. If so, output the current P. g (t) The algorithm terminates at the same time, otherwise jump to step 3-3 to continue execution.

[0036] Preferably, in the cluster collaborative trajectory planning method based on particle swarm beetle whiskers, the number of populations is 4, the number of particles in the population is 60, the maximum number of iterations is 3600 times, and the maximum allowed number of repetitions is 50 times.

[0037] Preferably, in the cluster collaborative trajectory planning method based on particle swarm beetle whiskers, c1=2.05, c2=2.05, w max =0.98, w min =0.4.

[0038] The beneficial effects of the present invention are as follows:

[0039] The method of the present invention avoids the local defects of the traditional PSO method and optimizes the optimal strike path for the UAV. The method also provides a valuable reference for cluster collaborative trajectory planning methods in other complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a simplified model diagram of the longicorn;

[0041] Figure 2 It is the trajectory planning diagram of the PSO algorithm;

[0042] Figure 3 It is the evolution curve of the cost function of the PSO algorithm;

[0043] Figure 4 It is the trajectory planning diagram of PSO_BAS algorithm;

[0044] Figure 5 It is the evolution curve of cost function of PSO_BAS algorithm;

[0045] Figure 6 It is the trajectory planning diagram of the PSO algorithm;

[0046] Figure 7 It is the evolution curve of the cost function of the PSO algorithm;

[0047] Figure 8 It is the trajectory planning diagram of PSO_BAS algorithm;

[0048] Fig. 9 It is the evolution curve of cost function of PSO_BAS algorithm;

[0049] Fig.10 is the weight change curve. DETAILED DESCRIPTION

[0050] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0051] The present invention proposes an improved particle swarm beetle whisker PSO_BAS method, which is divided into two levels: particle swarm optimization population and beetle whisker search algorithm re-optimization. At the first level, a particle swarm is used to optimize the entire particle population, and the algorithm's implicit parallel search strategy is used to complete the global search of the solution space; at the second level, the beetle whisker algorithm is used to re-optimize the optimization result of the previous level search to improve the algorithm accuracy. The beetle whisker algorithm relies on the fitness values ​​corresponding to the left and right whisker coordinates to compare and find a better search direction for the algorithm. As long as the algorithm step size is selected appropriately, it can meet good local optimization. In this way, the global optimal solution searched by the original particle swarm algorithm can be optimized again within a certain neighborhood range. If a better solution is found, the solution is used as the new global optimal solution. If a better solution is not found, the previous global optimal solution is still used.

[0052] 1. General description;

[0053] The optimization task allocation strategy combined with the BAS algorithm is adopted. Its overall structure is similar to the particle swarm optimization algorithm. The difference is that each time the algorithm is iterated, the global extreme value P of the particle is updated according to the particle swarm optimization algorithm. g (t) After the update, it does not directly enter the next round of optimization search, but first uses the beetle whisker update rule to update the global extreme value P at this time g (t) Perform optimal search within its neighborhood, and then enter the next round of optimal search. If a better global extreme value P is found within its neighborhood using the beetle whisker rule, g (t), then update it; if no better global extreme value P is found in the neighborhood g (t), the global extreme value P searched by the particle swarm is still maintained g (t). In this way, the PSO_BAS hybrid algorithm is to calculate the global extreme value P in each cycle of the particle swarm algorithm. g (t) The beetle whisker search step is added, thereby improving the local search capability of the algorithm.

[0054] 2. Methods and steps;

[0055] The optimization objective function is:

[0056]

[0057] Assume that the target search space is D-dimensional, the population size is N-dimensional, and the position vector of the i-th particle is recorded as X i =(x i1 ,x i2 ,...,x iD ), the velocity vector is denoted by V i =(v i1 ,v i2 ,...,v iD ), the historical optimal value of the i-th particle is recorded as pbest i =(p i1 ,p i2 ,...,p iD ), the global optimal value of the entire population in the target space is denoted as g best =(g1,g2,...,g D ). The particle position and velocity update formula is as follows:

[0058]

[0059] Among them, c1 represents the particle's own learning ability, c2 represents the social learning ability, r1 and r2 are uniform random numbers in the range of [0,1], t represents the current iteration number, i=1,2,...,N, N is the number of particles in the population, V i t denote the position and velocity of the t-th iteration respectively, V i t+1 They represent the position and speed of the t+1th iteration respectively. The inertia weight w represents the ability of the particle to maintain its original speed. The right side of (2) consists of three terms: the first term represents the particle inertia, indicating the tendency of the particle to maintain its previous speed; the second term represents the individual cognitive part, reflecting the particle's learning of its own historical experience, indicating that the particle is approaching its own historical best position; the third term represents the social cognitive part, reflecting the group historical experience of collaborative cooperation and information sharing among particles, indicating that the particle is approaching the group's optimal position.

[0060] The inertia weight w is adjusted using a linear decreasing strategy during the iteration process, and its calculation formula is as follows:

[0061] w=w max -t((w max -w min ) / t max ) (4)

[0062] Among them, G represents the maximum number of iterations, w max represents the maximum inertia weight; w min represents the minimum inertia weight; t represents the current iteration number.

[0063] The algorithm for the beetle whiskers is modeled as follows: X represents the centroid position of the beetle, X l Indicates the left whisker position, X r represents the position of the right whisker, d0 represents the distance between the two whiskers, and the simplified model of the longicorn is shown in Figure 1 shown.

[0064] From the description of the Longbeetle algorithm, we know that the direction of the Longbeetle's head is arbitrary during the optimization process, so it is represented by an arbitrary direction vector dir, and its normalized form is as follows:

[0065]

[0066] The formula for updating the optimal position of the longhorn beetle at time t is:

[0067]

[0068] Where f(·) represents the fitness function, X t , P t and P t-1 They represent the position of the longicorn at time t, the optimal position and the optimal position at time t-1 respectively.

[0069] The formula for updating the positions of the left and right whiskers of the longicorn at time t is:

[0070]

[0071] The fitness function value is calculated based on the left and right whiskers, and the fitness values ​​of the left and right whiskers are compared to determine the next movement direction of the longhorn beetle. The formula for updating the position of the longhorn beetle is:

[0072]

[0073] Wherein, step is the search step length, and step=c·d0, where c is a constant.

[0074] The specific steps are as follows:

[0075] Step 1: Initialize the parameters C1, C2, C, ω, Step, etc. in the algorithm, and randomly initialize the position X of each particle in the generated population i (t), calculate the fitness value corresponding to each particle according to the objective function;

[0076] Step 2: Let the optimal position of the individual at this time be P i (t) = X i(t), find the global optimal position P at this time g (t);

[0077] Step 3: Perform the following operations on each particle in the population: Update the velocity and position of the particle according to equations (2) and (3), and calculate the fitness value corresponding to the updated position; if the fitness value of the new position of the particle is better than the initial individual optimal value, update the individual optimal position and set P i (t) = X i (t); If the fitness value of the particle's new position is better than the global optimal value, then update the global optimal position, and let P g (t) = X i (t);

[0078] Step 4: Use the beetle whiskers to find the global optimal solution P for particles g (t) Perform local search and set the initial solution of the beetle to P = P g (t) and the optimal solution P gbest =P g (t);

[0079] Step 5: Determine whether the termination condition of the beetle search is met. If the termination condition is met, terminate P g (t) is the beetle whisker search process, and the optimization result P is output at the same time gbest , update P gbest =P g (t) and jump to Step 8; otherwise jump to the next step;

[0080] Step 6: Randomly generate the direction vector dir of the beetle's whiskers, and update the positions P of the left and right whiskers according to equations (7) and (8) left and P right ;

[0081] Step 7: According to the updated position P left and P right Calculate the corresponding fitness value. Update the position of the next beetle individual according to formula (9);

[0082] Step 8: Determine whether the algorithm has reached the iteration limit at this time. If it has, output the P at this time. g (t) The algorithm terminates at the same time, otherwise jump to Step 3 to continue execution.

[0083] Example:

[0084] 1. Experimental results;

[0085] 1.1 Initial setup of the brief environment;

[0086] The traditional PSO and improved PSO_BAS methods are used to plan the trajectory of the UAV strike process. The two methods use the same collaborative constraints and initial conditions such as the target information and environmental information of each UAV, as shown in Tables 1 and 2. The number of populations is 4, the number of particles in the population is 60, the maximum number of iterations is 3600, and the maximum number of repetitions allowed is 50. The learning factors c1 = 2.05, c2 = 2.05, and the inertia weight range is w max =0.98, w min =0.4.

[0087] Table 1. Collaborative constraints for trajectory planning

[0088]

[0089] Table 2 Initial information of trajectory planning

[0090]

[0091] The results of the traditional PSO method are as follows Figure 2 and Figure 3 As shown;

[0092] The results of the improved PSO_BAS method are as follows Figure 4 and Figure 5 As shown;

[0093] 1.2 Improve the initial setup of the environment;

[0094] In the above simulation, obstacles that the drone needs to avoid during flight are represented by spheres. However, in actual flight, missiles need to face multiple threats such as complex terrain, no-fly zones, and radar threats.

[0095] For complex terrain, the baseline terrain function is used to simulate the undulations of the terrain, and the mountain function is used to simulate the mountain threats in the flight environment.

[0096] The reference terrain function is shown in formula (10):

[0097]

[0098] The peak function is shown in formula (11):

[0099]

[0100] Among them, h i represents the height of the ith peak, (x 0i ,y 0i ) represents the center coordinates of the ith peak, (x si ,y si) represents the slope parameter of the ith peak along the x-axis and y-axis, and n represents the number of peaks.

[0101] In terms of the no-fly zone, it is equivalent to a cylinder, using (x i ,y i ) represents the center coordinates of the i-th cylinder, R i Indicates the radius of the cylinder, h i Indicates the height of the cylinder.

[0102] In terms of radar threat, it is equivalent to a hemisphere, using (x ri ,y ri ) represents the center of the i-th hemisphere, R ri Indicates the radius of the hemisphere. In addition, the minimum flight altitude of the drone must also be considered.

[0103] The added constraints are as follows:

[0104] No-fly zones and radar detection threats:

[0105] The path threat cost mainly includes radar detection threat and no-fly zone threat. According to the modeling of radar threat and no-fly zone threat in the previous section. Radar threat is a hemispherical area, and no-fly zone is a cylindrical area. The ratio of the distance between the path segment and the obstacle to the safe obstacle avoidance distance is used to describe the threat level, and the path threat constraint is:

[0106]

[0107] The first term in the above equation represents the radar detection threat, d ik1 represents the distance from the i-th path segment to the center of the k1-th threat hemisphere, r k1 represents the radius of the k1th threat hemisphere, d k10 represents the safe obstacle avoidance distance from the k1th threat hemisphere. The second term represents the no-fly zone threat, d ik2 represents the distance between the ith path segment and the center of the cylinder of equal height, r k2 represents the radius of the k2th threat cylinder, d k20 Represents the safe obstacle avoidance distance from the k2th threatening cylinder.

[0108] Minimum ground clearance constraint:

[0109] In order to ensure the safe flight of the drone during the mission, it is necessary to maintain a certain safety distance from the ground. Assuming the minimum height above the ground is H0, the minimum height above the ground constraint is:

[0110]

[0111] Where n represents the number of path points, H irepresents the terrain height at path point i, h i Represents the height of path point i.

[0112] The traditional PSO and improved PSO_BAS algorithms are used to plan the trajectory of the UAV strike process. The initial conditions such as the UAV information, target information and environmental information are shown in Table 3.

[0113] Table 3 Initial information of trajectory planning

[0114]

[0115] The constraints in path planning, such as minimum ground clearance, minimum turning radius, speed constraint, acceleration constraint, and path obstacle avoidance distance, are shown in Table 4.

[0116] Table 4. Constraints of trajectory planning

[0117]

[0118] Assume there are 4 drones, so there are 4 corresponding PSO_BAS populations, each population size is 30, learning factors c1 = 1.5, c2 = 1.5, inertia weight w max =0.9,w min =0.4, maximum number of iterations G=500, initial step size step=1, step size change factor eta=0.95, ratio of the beetle step size to the distance between the two whiskers c=5, algorithm weight λ=0.4.

[0119] The results of the traditional PSO method are as follows Figure 6 and Figure 7 ;

[0120] The results of the improved PSO_BAS method are as follows Figures 8 to 10 ;

[0121] 1.3 Experimental analysis;

[0122] 1.3.1 Brief environment;

[0123] From the above simulation, it can be seen that both the traditional PSO algorithm and the improved PSO_BAS algorithm can plan a strike path from the starting point to the target point for the UAV, and Figure 2 and Figure 4 It can be seen that the planned paths avoid obstacles and other threats, but it can also be seen from the figure that the path length using the PSO_BAS algorithm is shorter than that using the traditional PSO algorithm, indicating that the traditional PSO algorithm is not the optimal flight path and may fall into a local minimum. Figure 3 and Figure 5It can also be seen that the value of the stable convergence of the cost function using the traditional PSO algorithm is also larger than that using the improved PSO_BAS. This also confirms that the value obtained by using the traditional PSO algorithm is not the optimal value, while the improved PSO_BAS algorithm jumps out of the local minimum and finds the optimal attack path.

[0124] 1.3.2 Improve the environment

[0125] Figure 6 It means that the traditional PSO algorithm is used to plan the strike path from the starting point to the target point for each UAV. As can be seen from the figure, the particle swarm algorithm optimizes the strike flight path for each UAV and avoids the no-fly zone and radar threats during the flight. Figure 7 It represents the evolution curve of the cost function value with the number of iterations in the process of traditional PSO algorithm trajectory optimization. It can be seen from the figure that with the increase of iteration number, the cost function value continues to decrease and finally converges, indicating that the flight path is optimized for the UAV using the traditional PSO algorithm.

[0126] Depend on Figure 8 It can be seen that the improved particle swarm algorithm finds a flight path from the starting point to the target point for each UAV, avoids obstacles during the flight, and each path meets the constraint requirements. Fig. 9 The cost functions of the three paths change with the number of iterations during the trajectory planning process. As can be seen from the figure, as the number of iterations increases, the cost function value continues to decrease, and finally reaches a stable value and converges, indicating that an optimal strike path has been found for the drone. And compared with the cost function obtained by the traditional particle swarm algorithm Figure 7 Compared with the traditional PSO algorithm, the cost function value is smaller, indicating that the strike path length found for the UAV by the improved PSO_BAS algorithm is shorter. Under the same conditions, it saves fuel and reduces flight time, which is more in line with the expected planning requirements. Therefore, the optimal strike path obtained by the traditional PSO algorithm is not the optimal strike path, and it falls into the local minimum. The improved PSO_BAS algorithm avoids falling into the local minimum and finds the optimal strike path for the UAV, verifying the effectiveness of the improved PSO_BAS algorithm.

[0127] Fig.10This is the change of inertia weight w in the process of using the traditional PSO algorithm and the improved PSO_BAS algorithm. As can be seen from the figure, the inertia weight w shows a decreasing trend, and maintains a relatively large value in the early stage of the algorithm iteration, and the w value is smaller in the later stage of the algorithm. This can also make the traditional PSO algorithm and the improved PSO_BAS algorithm search globally with a larger speed step size in the early stage of the iteration, and search near the optimal point with a smaller w value in the later stage of the algorithm, so that the algorithm converges to the global optimum with a greater probability, and also avoids the algorithm from falling into the local minimum to a certain extent.

[0128] The present invention proposes an improved particle swarm beetle whiskers (PSO_BAS) method. Under simple environmental conditions and improved complex environmental conditions, both the traditional PSO particle swarm method and the improved PSO_BAS method optimize the flight path that satisfies the constraints for the UAV to strike the target process, but the improved PSO_BAS method well avoids the traditional PSO method's easy local defects and optimizes the optimal strike path for the UAV. This method also provides a valuable reference for cluster collaborative trajectory planning methods in other complex environments.

Claims

1. A cluster collaborative trajectory planning method based on particle swarm beetle whiskers, characterized in that: The steps include: Step 1: Set the optimization objective function as: Assume that the target search space is D-dimensional, the population size is N-dimensional, and the position vector of the i-th particle is recorded as X i =(x i1 ,x i2 ,...,x iD ), the velocity vector is denoted by V i =(v i1 ,v i2 ,...,v iD ), the historical optimal value of the i-th particle is recorded as pbest i =(p i1 ,p i2 ,...,p iD ), the global optimal value of the entire population in the target space is denoted as g best =(g1,g2,...,g D ); the particle position and velocity update formula are as follows: Among them, c1 represents the particle's own learning ability, c2 represents the social learning ability, r1 and r2 are uniform random numbers in the range of [0,1], t represents the current iteration number, i=1,2,...,N, N is the number of particles in the population, V i t denote the position and velocity of the t-th iteration respectively, V i t+1 They represent the position and velocity of the t+1th iteration respectively; the inertia weight w represents the ability of the particle to maintain its original velocity; The right side of formula (2) consists of three terms: the first term represents the particle inertia, indicating the tendency of the particle to maintain its previous speed; the second term represents the individual cognitive part, reflecting the particle's learning of its own historical experience, indicating that the particle is approaching its own historical optimal position; the third term represents the social cognitive part, reflecting the group historical experience of collaborative cooperation and information sharing among particles, indicating that the particle is approaching the group's optimal position; The inertia weight w is adjusted using a linear decreasing strategy during the iteration process, and its calculation formula is as follows: w=w max -t((w max -w min ) / t max ) (4) Among them, G represents the maximum number of iterations, w max represents the maximum inertia weight; w min represents the minimum inertia weight; t represents the current number of iterations; Step 2: The beetle whisker algorithm is modeled as follows: X represents the centroid position of the beetle, X l Indicates the left whisker position, X r represents the position of the right whisker, and d0 represents the distance between the two whiskers; The direction of the beetle's head is arbitrary during the optimization process, so it is represented by an arbitrary direction vector dir, and its normalized form is as follows: The formula for updating the optimal position of the longhorn beetle at time t is: Where f(·) represents the fitness function, X t , P t and P t-1 They represent the position of the longicorn at time t, the optimal position and the optimal position at time t-1 respectively; The formula for updating the positions of the left and right whiskers of the longicorn at time t is: The fitness function value is calculated based on the left and right whiskers, and the fitness values ​​of the left and right whiskers are compared to determine the next movement direction of the longhorn beetle; the formula for updating the position of the longhorn beetle is: Where step is the search step size, and step = c·d0, c is a constant; Step 3: Cluster collaborative trajectory planning method based on particle swarm beetle whiskers; Step 3-1: Initialize the parameters c1, c2, c, w, and Step in the algorithm, and randomly initialize the position X of each particle in the generated population. i (t), calculate the fitness value corresponding to each particle according to the objective function; Step 3-2: Let the optimal position of the individual at this time be P i (t) = X i (t), find the global optimal position P at this time g (t); Step 3-3: Perform the following operations on each particle in the population: Update the velocity and position of the particle according to equations (2) and (3), and calculate the fitness value corresponding to the updated position; if the fitness value of the new position of the particle is better than the initial individual optimal value, update the individual optimal position and set P i (t) = X i (t); If the fitness value of the particle's new position is better than the global optimal value, then update the global optimal position, and let P g (t) = X i (t); Step 3-4: Use the beetle whiskers to find the global optimal solution P for the particles g (t) Perform local search and set the initial solution of the beetle to P = P g (t) and the optimal solution P gbest =P g (t); Step 3-5: Determine whether the termination condition of the beetle search is met. If the termination condition is met, terminate P g (t) is the beetle whisker search process, and the optimization result P is output at the same time gbest , update P gbest =P g (t) and jump to step 3-8; otherwise, jump to the next step; Step 3-6: Randomly generate the direction vector dir of the beetle's whiskers, and update the positions P of the left and right whiskers according to equations (7) and (8) left and P right ; Step 3-7: According to the updated position P left and P right Calculate the corresponding fitness value; update the position of the next beetle individual according to formula (9); Step 3-8: Determine whether the iteration limit has been reached. If so, output the current P. g (t) The algorithm terminates at the same time, otherwise jump to step 3-3 to continue execution.

2. According to claim 1, a cluster collaborative trajectory planning method based on particle swarm beetle whiskers is characterized in that: In the cluster collaborative trajectory planning method based on particle swarm beetle whiskers, the number of populations is 4, the number of particles in the population is 60, the maximum number of iterations is 3600 times, and the maximum allowed number of repetitions is 50 times.

3. The method for cluster collaborative trajectory planning based on particle swarm beetle whiskers according to claim 1 is characterized in that: In the cluster collaborative trajectory planning method based on particle swarm beetle whiskers, c1=2.05, c2=2.05, w max =0.98, w min =0.4.

Images