Cluster formation control method, system, storage medium and electronic device based on brainstorming algorithm

Abstract

The present invention provides a cluster formation control method, system and storage medium based on a brainstorming algorithm, and relates to the technical field of cluster formation simulation. The method comprises: randomly generating N initial solutions in a solution space; performing cluster analysis on the N initial solutions to obtain m clusters, sorting the initial solutions included in the clusters according to the cost function value, and obtaining the cluster centers corresponding to the clusters; randomly selecting a cluster center, and in the case of obtaining a chaotic solution sequence, sorting the chaotic solutions included in the chaotic solution sequence according to the cost function value to obtain the optimal chaotic solution, and replacing the randomly selected cluster center with the optimal chaotic solution; based on all cluster centers, generating an auxiliary solution corresponding to each initial solution; updating the initial solution based on each auxiliary solution to obtain a new solution for each initial solution; sorting all new solutions according to the cost function value, and outputting the optimal new solution; configuring the optimal new solution as the control amount of the cluster formation, and updating the state of the cluster formation.

Description

Technical Field

[0001] The present invention relates to the technical field of cluster formation simulation, and in particular to a cluster formation control method, system and storage medium based on a brainstorming algorithm. Background Art

[0002] In cluster control problems, it is often necessary to solve the control quantity, which can be solved by swarm intelligence algorithms. The Brain Storm Optimization algorithm comes from brainstorming meetings, where participants can imagine freely, generate a group of ideas in each round, and classify and sort the ideas. Participants refer to, supplement and improve the existing ideas, inspire and motivate each other, and finally get the discussion results.

[0003] The original BSO uses k-means clustering. The cluster center is selected by sorting the cost function within the cluster, and a solution is randomly selected to generate an auxiliary solution using probability selection. The new solution is used to calculate the new solution of each solution. The calculation of the new solution is based on the auxiliary solution and the random value of the normal distribution. Then, based on the cost function value, decide whether to use the new solution to update the original solution. There are two main ways for the original BSO to generate auxiliary solutions. One is to select one cluster to generate, and the other is to select two clusters to generate. These two generation methods are determined by probability. p1 is the probability of randomly selecting a cluster, and p2=1-p1 is the probability of randomly selecting two clusters. When selecting a cluster center, p 1c is the probability of selecting a cluster center as an auxiliary solution, otherwise a non-cluster center is selected as an auxiliary solution. In the case of selecting two cluster centers, p 2c is the probability of selecting two cluster centers to generate auxiliary solutions, otherwise two non-cluster centers are selected to generate auxiliary solutions, and w1 and w2 are the weight coefficients of the two solutions when the auxiliary solutions are generated.

[0004] There are two main types of algorithms for improving BSO. One is to change the clustering method and replace the original k-means clustering with other clustering methods. The other is to change the formula for updating the solution. Some introduce the predator and prey mechanism and then perform chaotic search on each updated solution. Some make the random one-dimensional solution in a randomly selected cluster chaotic. Some use multi-branch chaotic mutation to generate a chaotic mutation operator, and perform chaotic search around when the solution is updated. Some adjust the probability parameters of the grouping strategy and improve the variance of the Gaussian random number.

[0005] Existing improvement methods mainly focus on clustering methods and solution updating methods, and pay less attention to the solution replacement mechanism and auxiliary solution generation mechanism in the BSO algorithm. Summary of the invention

[0006] The purpose of the present invention is to provide a cluster formation control method, system and storage medium based on the brainstorming algorithm. Without changing the clustering method, the solution replacement mechanism is changed, a chaotic map is selected to generate a chaotic sequence, a chaotic solution is generated for replacing the solution, all cluster centers are considered and an auxiliary solution is generated in combination with a chaotic operator, so as to iteratively update the solution.

[0007] In a first aspect, the present invention provides a cluster formation control method based on a brainstorming algorithm, comprising the following steps:

[0008] S10, defining a solution space, the solution space having an upper limit of the solution space and a lower limit of the solution space, and randomly generating N initial solutions in the solution space, N>1;

[0009] S11, performing cluster analysis on the N initial solutions to obtain m clusters, where m≤N, sorting the initial solutions included in each cluster according to the cost function value to obtain the cluster center corresponding to each cluster;

[0010] S12, when the preset conditions are met, randomly selecting a cluster center, and when a chaotic solution sequence is obtained, sorting the chaotic solutions included in the chaotic solution sequence according to the cost function value to obtain an optimal chaotic solution, and replacing the randomly selected cluster center with the optimal chaotic solution;

[0011] S13, based on all cluster centers, generating an auxiliary solution corresponding to each initial solution;

[0012] S14, updating the initial solution based on each auxiliary solution to obtain a new solution for each initial solution;

[0013] S15. When it is determined that the maximum number of iterations reaches the preset number of iterations, all new solutions are sorted according to the cost function values, and the optimal new solution is output;

[0014] S16. Configure the optimal new solution as the control amount of the cluster formation and update the state of the cluster formation.

[0015] Under the condition of adopting the above technical scheme, the present invention improves the solution replacement mechanism in the original BSO algorithm, selects chaotic mapping, generates chaotic sequence and chaotic solution sequence, and selects the optimal chaotic solution to replace the solution. By utilizing the characteristic that chaotic mapping is sensitive to input, the chaos of the replacement solution strengthens the global search ability of the algorithm, so that the area where the optimal solution is located can be found more quickly. The present invention improves the auxiliary solution generation mechanism in the original BSO algorithm. The original algorithm selects one cluster or two clusters to generate auxiliary solutions, and cannot comprehensively consider all clusters. The present invention uses a chaotic sequence to generate a set of chaotic weight coefficients, which are generated in combination with each cluster center to ensure a certain global search ability and strengthen the local search ability, so as to find better results.

[0016] Optionally, after S14, the control method of the cluster formation based on the brainstorming algorithm further includes:

[0017] When it is determined that the maximum number of iterations has not reached the preset number of iterations, S11 to S14 are executed in a loop until the maximum number of iterations reaches the preset number of iterations, so as to output an optimal new solution.

[0018] Optionally, the upper and lower limits of the solution space are represented by X max and X min ;

[0019] Each initial solution can be expressed as

[0020] Where D is the dimension of the solution space, is the position of the initial solution in the solution space, and is also the value of the initial solution in each dimension.

[0021] Optionally, the N initial solutions are clustered using the k-means clustering method to obtain m clusters, which are denoted as C j ,j∈{1,2,3,...,m};

[0022] The cluster center is the optimal initial solution in each cluster sorted by cost function value. The cluster center is represented by X Cj .

[0023] Optionally, step S12 includes:

[0024] Determine the probability p r , randomly generate a random number p between [0,1] r0 , if p r0 <p r , randomly select a cluster center where j r ∈{1,2,3,...,m} is a randomly selected cluster number;

[0025] Randomly generate a D-dimensional vector cr(0) as the seed of the sequence and select a chaotic map f rmap , perform n r The chaotic sequence {cr(1),cr(2),...,cr(n r )}.

[0026] cr(i r ) = f rmap (cr(i r -1)),i r =1,2,...,n r (6)

[0027] The corresponding chaotic solution sequence obtained by using formula (2) is {cx(1),cx(2),...,cx(n r )}

[0028] cx(i r )=cr(i r )(X max -X min )+X min (7)

[0029] Calculate the cost function value of each chaotic solution. The chaotic solution with the best cost function value is cx best Replace the selected cluster center

[0030] Optionally, in S13,

[0031] For each initial solution X i Generate an auxiliary solution X ih ;

[0032] The generation of auxiliary solutions considers all cluster centers Randomly generate an m-dimensional vector ch(0) = [ch1(0), ch2(0), ..., ch m (0)] as the seed of the sequence, and select a chaotic map f hmap , perform n h The chaotic weight coefficient sequence {ch(1), ch(2), ..., ch(n) is generated by using formula (3) h )}.

[0033] ch(i h ) = f hmap (cr(i h -1)),i h =1,2,...,n h (8)

[0034] Use ch(n h )=[ch1(n h ),ch2(n h ),...,ch m (n h )] and cluster centers to generate auxiliary solutions X ih ;

[0035]

[0036] Optionally, in step S14,

[0037] Based on each auxiliary solution X ih For the initial solution X iGenerate a new solution X inew ;

[0038] The update formula is as follows

[0039] X inew =X ih +logsig((0.5×T max -T now ) / k)×X rand ×norm(0,1,D) (10)

[0040] logsig() is a sigmoid function with an output between (0,1). k is a parameter that affects the slope of the function. max is the maximum number of iterations, T now is the current iteration number, norm(0,1,D) generates a D-dimensional vector of normally distributed random numbers;

[0041] Calculate X inew The cost function value is, if it is higher than X i The cost function value is better, update the alternative solution X i =X inew .

[0042] In a second aspect, the present invention further provides a cluster formation control system based on a brainstorming algorithm, comprising:

[0043] An initial solution determination module, in which a solution space is defined, the solution space has an upper limit of a solution space and a lower limit of a solution space, and N initial solutions are randomly generated in the solution space, where N>1;

[0044] Clustering module: The clustering module performs cluster analysis on N initial solutions to obtain m clusters, where m≤N. The initial solutions included in each cluster are sorted according to the cost function value to obtain the cluster center corresponding to each cluster.

[0045] The optimal chaotic solution replacement module randomly selects a cluster center when the preset conditions are met, and when a chaotic solution sequence is obtained, the chaotic solutions included in the chaotic solution sequence are sorted according to the cost function value to obtain the optimal chaotic solution, and the randomly selected cluster center is replaced with the optimal chaotic solution;

[0046] The auxiliary solution determination module generates an auxiliary solution for each initial solution based on all cluster centers;

[0047] An updating module, which updates the initial solution based on each auxiliary solution to obtain a new solution for each initial solution;

[0048] The optimal new solution determination module, when determining that the maximum number of iterations reaches the preset number of iterations, sorts all new solutions according to the cost function value, outputs the optimal new solution, configures the optimal new solution as the control amount of the cluster formation, and updates the state of the cluster formation.

[0049] In a third aspect, the present invention further provides a readable storage medium, which is used to store program code. When the program code is read and executed by a computer, it executes the cluster formation control method based on the first aspect and / or any optional brainstorming algorithm.

[0050] In a fourth aspect, the present invention also provides an electronic device comprising: one or more processors; and one or more machine-readable media having instructions stored thereon, which, when executed by the one or more processors, enables execution of the first aspect and / or any optional brainstorming algorithm-based cluster formation control method. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0052] Figure 1 An overview diagram of a cluster formation control method based on a brainstorming algorithm provided in an embodiment of the present invention;

[0053] Figure 2 A flowchart of a cluster formation control method based on a brainstorming algorithm provided in an embodiment of the present invention;

[0054] Figure 3 A comparison diagram of the optimal new solution obtained by the prior art and the optimal new solution in the control method provided by the embodiment of the present invention;

[0055] Figure 4 A graph showing the change of the cost function in the prior art and the control method provided in the embodiment of the present invention;

[0056] Figure 5 A trajectory diagram of a drone formation when the control method provided by an embodiment of the present invention is applied to drones;

[0057] Figure 6 A diagram of drone state changes when the control method provided by an embodiment of the present invention is applied to a drone. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Figure 1 and Figure 2 The overview diagram and flow chart of the cluster formation control method based on the brainstorming algorithm provided in the embodiment of the present invention are respectively illustrated, see Figure 1 and Figure 2 , a cluster formation control method based on a brainstorming algorithm provided by an embodiment of the present invention comprises the following steps:

[0060] S10. Define a solution space, where the solution space has an upper limit and a lower limit, and randomly generate N initial solutions in the solution space, where N>1.

[0061] In this step, the upper limit and lower limit of the solution space are represented by X max and X min ;

[0062] Each initial solution can be expressed as

[0063] Where D is the dimension of the solution space, is the position of the initial solution in the solution space, and is also the value of the initial solution in each dimension.

[0064] S11. Perform cluster analysis on the N initial solutions to obtain m clusters, where m≤N. Sort the initial solutions included in each cluster according to the cost function value to obtain the cluster center corresponding to each cluster.

[0065] In this step, the k-means clustering method is used to perform cluster analysis on the N initial solutions to obtain m clusters, which are represented by C j , j∈{1, 2, 3,..., m};

[0066] The cluster center is the optimal initial solution in each cluster sorted by cost function value. The cluster center is represented by X Ci .

[0067] S12. When the preset conditions are met, randomly select one of the cluster centers. When a chaotic solution sequence is obtained, sort the chaotic solutions included in the chaotic solution sequence according to the cost function value to obtain an optimal chaotic solution, and replace the randomly selected cluster center with the optimal chaotic solution.

[0068] In this step, the probability p is determined r , randomly generate a random number p between [0, 1] r0 , if p r0 <p r , randomly select a cluster center where j r ∈{1, 2, 3, ..., m} is a randomly selected cluster number;

[0069] Randomly generate a D-dimensional vector cr(0) as the seed of the sequence and select a chaotic map f rmap , perform n r The chaotic sequence {cr(1), cr(2), ..., cr(n) is generated by using formula (1). r )}.

[0070] cr(i r ) = f rmap (cr(i r -1)),i r =1,2,...,n r (11)

[0071] The corresponding chaotic solution sequence obtained by formula (2) is {cx(1), cx(2), ..., cx(n r )}

[0072] cx(i r )=cr(i r )(Xminmin max (12)

[0073] Calculate the cost function value of each chaotic solution. The chaotic solution with the best cost function value is cx best Replace the selected cluster center

[0074] S13. Based on all the cluster centers, generate an auxiliary solution corresponding to each of the initial solutions.

[0075] In this step, for each initial solution X i Generate an auxiliary solution X ih ;

[0076] The generation of auxiliary solutions considers all cluster centers Randomly generate an m-dimensional vector ch(0) = [ch1(0), ch2(0), ..., ch m (0)] as the seed of the sequence, and select a chaotic map f hmap , perform n hThe chaotic weight coefficient sequence {ch(1), ch(2), ..., ch(n) is generated by using formula (3) h )}.

[0077] ch(i h ) = f hmap (cr(i h -1)),i h =1,2,...,n h (13)

[0078] Use ch(n h )=[ch1(n h ), ch2(n h ), ..., ch m (n h )] and cluster centers to generate auxiliary solutions X ih ;

[0079]

[0080] S14. Update the initial solution based on each of the auxiliary solutions to obtain a new solution for each of the initial solutions.

[0081] Based on each auxiliary solution X ih For the initial solution X i Generate a new solution X inew ;

[0082] The update formula is as follows

[0083] X inew =X ih +logsig((0.5×T max -T now ) / k)×X rand ×norm(0,1,D) (15)

[0084] logsig() is a sigmoid function, the output is between (0, 1), k is a parameter that affects the slope of the function, T max is the maximum number of iterations, T now is the current iteration number, norm(0, 1, D) generates a D-dimensional vector of normally distributed random numbers;

[0085] Calculate X inew The cost function value is, if it is higher than X i The cost function value is better, update the alternative solution X i =X inew .

[0086] S15. When it is determined that the maximum number of iterations reaches the preset number of iterations, all the new solutions are sorted according to the cost function values, and the optimal new solution is output.

[0087] S16, configuring the optimal new solution as the control amount of the cluster formation, and updating the state of the cluster formation. The cluster formation may be a drone formation.

[0088] As a possible implementation method, after S14, the control method of the cluster formation based on the brainstorming algorithm also includes:

[0089] When it is determined that the maximum number of iterations has not reached the preset number of iterations, S11 to S14 are executed in a loop until the maximum number of iterations reaches the preset number of iterations, so as to output an optimal new solution.

[0090] In a second aspect, an embodiment of the present invention further provides a cluster formation control system based on a brainstorming algorithm, comprising:

[0091] An initial solution determination module, in which a solution space is defined, the solution space has an upper limit of a solution space and a lower limit of a solution space, and N initial solutions are randomly generated in the solution space, where N>1;

[0092] Clustering module: The clustering module performs cluster analysis on N initial solutions to obtain m clusters, where m≤N. The initial solutions included in each cluster are sorted according to the cost function value to obtain the cluster center corresponding to each cluster.

[0093] The optimal chaotic solution replacement module randomly selects a cluster center when the preset conditions are met, and when a chaotic solution sequence is obtained, the chaotic solutions included in the chaotic solution sequence are sorted according to the cost function value to obtain the optimal chaotic solution, and the randomly selected cluster center is replaced with the optimal chaotic solution;

[0094] The auxiliary solution determination module generates an auxiliary solution for each initial solution based on all cluster centers;

[0095] An updating module, which updates the initial solution based on each auxiliary solution to obtain a new solution for each initial solution;

[0096] The optimal new solution determination module, when determining that the maximum number of iterations reaches the preset number of iterations, sorts all new solutions according to the cost function value, outputs the optimal new solution, configures the optimal new solution as the control amount of the cluster formation, and updates the status of the cluster formation.

[0097] In a third aspect, the present invention further provides a readable storage medium, which is used to store program code. When the program code is read and executed by a computer, it executes the cluster formation control method based on the first aspect and / or any optional brainstorming algorithm.

[0098] In a fourth aspect, an embodiment of the present invention further provides an electronic device comprising: one or more processors; and one or more machine-readable media having instructions stored thereon, which, when executed by the one or more processors, enables execution of the first aspect and / or any optional brainstorming algorithm-based cluster formation control method.

[0099] In order to understand the above technical solution of the present invention, the above technical solution of the present invention is described in detail below through specific embodiments to verify the effectiveness of the cluster formation control method proposed in the present invention.

[0100] Embodiment 1:

[0101] Function Find the optimal solution, where x=[x1,x2,...x 30 ]. The specific implementation steps are as follows:

[0102] The parameters of the original brainstorming algorithm are designed as follows: the number of individuals N = 200, the maximum number of iterations T max =200, the number of clusters m = 5, the probability p in the replacement mechanism r =0., the probability of generating auxiliary solutions in the mechanism is p1 = 0.5, p2 = 0.5, p 1c =0.3, p 2c =0.3, the coefficients of the auxiliary solution w1=0.3, w2=0.4, in the formula for iteratively updating the solution, the slope coefficient k=20. The dimension of the solution space D=30, and the upper and lower limits of the solution space in each dimension are [-100, 100].

[0103] The parameter design in the improved brainstorming algorithm is the same as that in the original brainstorming algorithm. However, in the solution replacement and auxiliary solution generation mechanism, the two chaotic maps f rmap and f hmap The chaotic mapping is selected as Logistic mapping, f rmap (cc) = f hmap (cc)=4×cc×(1-cc), 0<cc<1.

[0104] The optimal solutions obtained by the original BSO and the improved BSO are as follows: Figure 3 As shown in the figure, it can be seen that the improved BSO algorithm is closer to the actual optimal solution 0. The cost function value corresponding to the optimal solution obtained by the original BSO is 0.3722, and the cost function value corresponding to the optimal solution obtained by the improved BSO is 0.0471, which is better than the original BSO. The cost function changes of the original BSO and the improved BSO are shown in Figure 4 As shown, it can be seen that the improved BSO algorithm has better optimization results and faster convergence speed, which verifies the feasibility and effectiveness of the present invention.

[0105] Embodiment 2:

[0106] The solution of UAV swarm formation control is to take three UAVs as an example. In a plane at a certain height, the UAVs are regarded as particles. Their initial states are as follows: x and y are the position coordinates of the UAVs, v is the linear velocity of the UAV, and θ is the yaw angle of the UAV. When solving the control quantity for each UAV in the swarm, the solution space dimension D = 2, and the linear velocity acceleration and yaw angular velocity are solved respectively. The acceleration range is [-7m / s 2 ,+7m / s 2 ], the angular velocity range is [-0.25rad / s, +0.25rad / s]. A numerical simulation of three UAVs forming a triangle formation is performed, and the trajectory of the UAVs is shown in the figure Figure 5 As shown in the figure, it can be seen that the drones can form a triangle formation and run according to the predetermined trajectory. The state changes of each drone are as follows Figure 6 As shown in the figure, it can be seen that the improved BSO algorithm can solve the control quantity of each UAV within the range, and the speed and angle of each UAV in the formation tend to be consistent. It is verified that the improved BSO algorithm can be used to solve cluster control problems.

[0107] Table 1 Initial state of the drone

[0108]

[0109] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0110] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A cluster formation control method based on brainstorming algorithm, characterized in that: The following steps are involved: S10, defining a solution space of the acceleration of the linear velocity and the yaw angular velocity of each UAV in the cluster, wherein the solution space has an upper limit and a lower limit of the solution space, and randomly generating N initial solutions of the acceleration of the linear velocity and the yaw angular velocity in the solution space, where N>1; S11, performing cluster analysis on the N initial solutions to obtain m clusters, where m≤N, and sorting the initial solutions included in each cluster according to the cost function value to obtain the cluster center corresponding to each cluster; S12, when the preset conditions are met, randomly selecting one of the cluster centers, and when a chaotic solution sequence of the acceleration of the linear velocity and the yaw angular velocity is obtained, sorting the chaotic solutions included in the chaotic solution sequence according to the cost function value, obtaining the optimal chaotic solution of the acceleration of the linear velocity and the yaw angular velocity, and replacing the randomly selected cluster center with the optimal chaotic solution; S13, based on all the cluster centers, generating an auxiliary solution of the acceleration of the linear velocity and the yaw angular velocity corresponding to each of the initial solutions; S14, updating the initial solution based on each auxiliary solution to obtain a new solution of the acceleration of the linear velocity and the yaw angular velocity corresponding to each of the initial solutions; S15. When it is determined that the maximum number of iterations reaches the preset number of iterations, all the new solutions are sorted according to the cost function values, and the optimal new solutions of the acceleration of the linear velocity and the yaw angular velocity are output; S16, configuring the optimal new solution as the control amount of the cluster formation, and updating the state of the cluster formation; in, In said S13, Corresponding to each of the initial solutions X i Generate an auxiliary solution X of the acceleration and yaw angular velocity of the linear velocity ih ,include: The generation of auxiliary solutions considers all cluster centers Randomly generate an m-dimensional vector ch(0) = [ch1(0), ch2(0), ..., ch m (0)] as the seed of the sequence, and select a chaotic map f hmap , perform n h The chaotic weight coefficient sequence {ch(1), ch(2), ..., ch(n) is generated by using formula (3) h )}, ch(i h )=f hmap (cr(i h -1)),i h =1,2,...,n h (3) Use ch(n h )=[ch1(n h ),ch2(n h ),...,ch m (n h )] and the cluster center according to formula (4) to generate the auxiliary solution X ih , 2. The cluster formation control method based on the brainstorming algorithm according to claim 1 is characterized in that: After S14, the control method of cluster formation based on the brainstorming algorithm further includes: When it is determined that the maximum number of iterations has not reached the preset number of iterations, S11 to S14 are executed in a loop until the maximum number of iterations reaches the preset number of iterations, so as to output the optimal new solution.

3. The cluster formation control method based on the brainstorming algorithm according to claim 1 is characterized in that: The upper limit of the solution space and the lower limit of the solution space are respectively represented by X max and X min ; Each of the initial solutions can be expressed as i∈{1,2,3,...,N}; Where D is the dimension of the solution space, is the position of the initial solution in the solution space, and is also the value of the initial solution in each dimension.

4. The cluster formation control method based on the brainstorming algorithm according to claim 3 is characterized in that: The N initial solutions are clustered using the k-means clustering method to obtain m clusters, which are represented by C j ,j∈{1,2,3,...,m}; The cluster center is the optimal initial solution obtained by sorting the cost function values ​​in each cluster, and the cluster center is represented by X Cj .

5. The cluster formation control method based on the brainstorming algorithm according to claim 4 is characterized in that: The step S12 comprises: Determine the probability p r , randomly generate a random number p between [0,1] r0 , if p r0 <p r , randomly select one of the cluster centers where j r ∈{1,2,3,...,m} is a randomly selected cluster number; Randomly generate a D-dimensional vector cr(0) as the seed of the sequence and select a chaotic map f rmap , perform n r The chaotic sequence {cr(1),cr(2),...,cr(n r )}; cr(i r )=f rmap (cr(i r -1)),i r =1,2,...,n r (1) The corresponding chaotic solution sequence obtained by using formula (2) is {cx(1),cx(2),...,cx(n r )} cx(i r )=cr(i r )(X max -X min )+X min (2) Calculate the cost function value of each chaotic solution. The chaotic solution with the best cost function value is cx best Replace the selected cluster center 6. The cluster formation control method based on the brainstorming algorithm according to claim 1 is characterized in that: In step S14, Based on each of the auxiliary solutions X ih For the initial solution X i Update to obtain the corresponding initial solution X i New solution of linear velocity acceleration and yaw angular velocity X inew , The update formula is as follows X inew =X ih +logsig((0.5×T max -T now ) / k)×X rand ×norm(0,1,D) (5) logsig() is a sigmoid function with an output between (0,1). k is a parameter that affects the slope of the function. max is the maximum number of iterations, T now is the current iteration number, norm(0,1,D) generates a D-dimensional vector of normally distributed random numbers; Calculate X inew The cost function value is, if it is higher than X i The cost function value is better, update the alternative solution X i =X inew .

7. A cluster formation control system based on brainstorming algorithm, characterized in that: include: An initial solution determination module, in which a solution space of the acceleration of the linear velocity and the yaw angular velocity of each UAV in the cluster is defined, wherein the solution space has an upper limit of the solution space and a lower limit of the solution space, and N initial solutions of the acceleration of the linear velocity and the yaw angular velocity are randomly generated in the solution space, where N>1; A clustering module, wherein the clustering module performs cluster analysis on the N initial solutions to obtain m clusters, where m≤N, and sorts the initial solutions included in each cluster according to the cost function value to obtain a cluster center corresponding to each cluster; The optimal chaotic solution replacement module randomly selects one of the cluster centers when a preset condition is met, and when a chaotic solution sequence of the acceleration of the linear velocity and the yaw angular velocity is obtained, the chaotic solutions included in the chaotic solution sequence are sorted according to the cost function value to obtain the optimal chaotic solution of the acceleration of the linear velocity and the yaw angular velocity, and the randomly selected cluster center is replaced with the optimal chaotic solution; The auxiliary solution determination module generates an auxiliary solution of the acceleration of the linear velocity and the yaw angular velocity corresponding to each of the initial solutions based on all the cluster centers. in, For each of the initial solutions X i Generate an auxiliary solution X ih , The generation of auxiliary solutions considers all cluster centers {X C1 ,X C2 ,...,X Cm }, randomly generate an m-dimensional vector ch(0) = [ch1(0), ch2(0), ..., ch m (0)] as the seed of the sequence, and select a chaotic map f hmap , perform n h The chaotic weight coefficient sequence {ch(1), ch(2), ..., ch(n) is generated by using formula (3) h )}, ch(i h )=f hmap (cr(i h -1)),i h =1,2,...,n h (1) Use ch(n h )=[ch1(n h ),ch2(n h ),...,ch m (n h )] and cluster centers to generate auxiliary solutions X ih , An updating module, which updates the initial solution based on each auxiliary solution to obtain a new solution of the acceleration of the linear velocity and the yaw angular velocity corresponding to each initial solution; The optimal new solution determination module, when determining that the maximum number of iterations reaches a preset number of iterations, sorts all the new solutions according to the cost function value, and outputs the optimal new solution of the acceleration of the linear velocity and the yaw angular velocity.

8. A readable storage medium, characterized in that: The readable storage medium is used to store program codes, and when the program codes are read and executed by a computer, the cluster formation control method based on the brainstorming algorithm as described in any one of claims 1 to 6 is executed.

9. An electronic device, characterized in that: include: one or more processors; and one or more machine-readable media having instructions stored thereon, which, when executed by the one or more processors, enable the cluster formation control method based on the brainstorming algorithm as described in any one of claims 1 to 6 to be executed.

Images