A swarm intelligence optimization method based on grid governance
Patent Information
- Application Number
- CN202410362995.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-28
- Publication Date
- 2026-09-01
- Estimated Expiration
- 2044-03-28
AI Technical Summary
[0002]智能优化算法广泛应用于连续函数的极值求解以及离散的组合优化问题中，传统的粒子群、蚁群、模拟退火等智能优化算法主要模拟动物行为或自然过程，缺少借鉴人类群体活动经验，算法过程不易于理解，且普遍存在多峰值高维函数极值求解过程慢、易陷入局部最优、超参数设置复杂等问题，因此，研究基于人类群体活动的智能优化方法具有重要的应用价值
[0042] (1) The process of condensing community grid governance experience into a swarm intelligence algorithm can effectively solve the problem of finding the extreme value of complex functions;
Smart Images
Figure CN118297096B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent optimization technology, specifically to a swarm intelligence optimization method based on grid governance. Background Technology
[0002] Intelligent optimization algorithms are widely used in solving the extrema of continuous functions and discrete combinatorial optimization problems. Traditional intelligent optimization algorithms such as particle swarm optimization, ant colony optimization, and simulated annealing mainly simulate animal behavior or natural processes, lacking the experience of human group activities. The algorithm process is not easy to understand, and they generally suffer from problems such as slow extrema solving process of multi-peak high-dimensional functions, easy getting trapped in local optima, and complex hyperparameter settings. Therefore, researching intelligent optimization methods based on human group activities has important application value. Summary of the Invention
[0003] Purpose of the invention: The purpose of this invention is to provide a swarm intelligence optimization method based on grid governance.
[0004] Technical solution: The present invention provides a swarm intelligence optimization method based on grid governance, comprising the following steps:
[0005] S1. Divide the region of the intelligent optimization group into grids; including defining hyperparameters, initializing grid member positions, evaluating grid member fitness, calculating the number of grid members, and generating grid members;
[0006] S2. Update the positions of grid members to achieve grid evolution; this includes: calculating the fitness of grid members within each grid and selecting the one with the best fitness as the optimal grid member. The optimal members of each grid are ranked by fitness to obtain the globally optimal grid members. Select the currently globally optimal grid member. As role models, all grid members evolve to obtain new grid member positions based on the difference between their own positions and the role model's positions.
[0007] S3. Select new grid members; this includes: calculating the fitness of each grid member evolved in step S2; sorting the grid members according to their fitness values, and selecting the grid member with the best fitness and whose distance in each dimension is not less than the offset distance between the grid member and the grid manager. of Each grid member will be considered as a candidate for the next term of grid members. The number of grid members is determined. The iteration termination condition is checked. If the upper limit of the number of iterations is reached, or the set optimal fitness threshold is met, the iteration ends and the current optimal grid member position and fitness value are output. If the termination condition is not met, the candidate grid member is selected as the grid member of the next cycle and the iteration is returned to step S1 to recalculate the number of grid members. The iteration continues until the termination condition is met.
[0008] Furthermore, step S1 specifically includes:
[0009] (1) Define hyperparameters: Hyperparameters include the number of grid cells. Grid standard member number Maximum number of iterations Offset distance between grid members and grid operators , probability of mutation of grid members and grid member learning rate ;
[0010] (2) Initialize grid worker positions: Each grid corresponds to one grid worker, and the grid worker is initialized using a uniformly distributed random number. The position vector of a grid worker ,in For the location variable dimension, the grid worker Represents a feasible solution to an optimization problem, each of its elements All belong to the bounded vectors under the constraints. and upper limit vector Values within the range, of which , , That is:
[0011]
[0012]
[0013] in, Representative generated in Within the constraint range, it follows a uniform distribution. A number of random numbers, the same as the number of grid cells;
[0014] (3) Evaluate the fitness of grid workers: Calculate the fitness of each grid worker according to the form of the objective function to be solved. ;
[0015] (4) Calculate the number of grid members: Determine the number of grid members based on the grid member fitness. ;
[0016] (5) Generate mesh members: Mesh members are divided into two categories: regular members and mutated members. Regular members are generated by the position offset of the mesh operator, based on a preset offset distance. The calculation shows that before each generation of grid members, a random number is first generated, and it is determined that the random number is lower than the member mutation probability value. When the time is right, a mutated member is generated. The mutated member is not affected by the position of the grid member. Only the position variable that meets the constraints is randomly generated. When the random number is greater than the mutation probability, a regular member is generated.
[0017] Furthermore, the objective to be solved in step (3) is the airspace coverage of the radar at each altitude layer in the multi-radar network detection system, the optimization variable is the radar position, and the model to be optimized is:
[0018]
[0019] in, This represents a feasible solution that satisfies the constraints. For the first Location information of radar deployment. , The first The radar deployment provides longitude, latitude, and altitude information. , The scope of the deployable area is defined by constraints. These represent the upper limits for the longitude, latitude, and altitude information of the deployable area, respectively. These are the lower limits for the longitude, latitude, and altitude information of the deployable area, respectively. Represents the radar under current deployment for the first Spatial coverage obtained after discretizing each height layer Represents the number of airspace height layers. This is the minimum airspace coverage requirement.
[0020] Furthermore, the method for determining the number of grid members in step (4) is as follows:
[0021]
[0022] in For the standard number of members in the grid, The minimum fitness among all grid members, For maximum fitness, For the first Each grid worker's fitness value Extremely small numbers introduced to prevent division by zero. To limit the maximum number of grid members, This indicates the rounding operation.
[0023] Furthermore, in step (5), the regular members are generated by the grid operator's position offset, based on a preset offset distance. The calculation yields the following results:
[0024]
[0025] in, Representing the The first grid A regular member, A random number between 0 and 1, where 'a' is a randomly generated value of 0, 1, or -1. This is the preset offset distance between grid members and grid operators.
[0026] Furthermore, the method for calculating the position of the mutated member in step (5) is as follows:
[0027]
[0028] in, For the location variable dimension, A random number between 0 and 1 For the first The first grid A variant member.
[0029] Furthermore, in step S2, all grid members evolve to obtain new grid member positions based on the difference between their own positions and the leader's position. The update formula is as follows:
[0030]
[0031] in, Representing the evolved first The first grid Each grid member, For the first The first grid Each grid member, The preset member learning rate, A random number between 0 and 1.
[0032] The present invention provides a swarm intelligence optimization system based on grid governance, comprising:
[0033] The grid partitioning unit is used to partition the region of the intelligent optimization population into a grid; it includes defining hyperparameters, initializing grid member positions, evaluating grid member fitness, calculating the number of grid members, and generating grid members.
[0034] The grid evolution unit is used to update the positions of grid members to achieve grid evolution; it includes: calculating the fitness of grid members within each grid and selecting the one with the best fitness as the optimal member of the grid. The optimal members of each grid are ranked by fitness to obtain the globally optimal grid members. Select the currently globally optimal grid member. As role models, all grid members evolve to obtain new grid member positions based on the difference between their own positions and the role model's positions.
[0035] The grid leadership transition unit is used for selecting new grid members. This includes: calculating the fitness of each evolved grid member in the grid evolution unit; ranking the grid members based on their fitness values; and selecting the grid member with the best fitness and whose positional variables in each dimension are not less than the offset distance between the grid member and the original grid member. of Each grid member will be considered as a candidate for the next term of grid members. The number of grid members is denoted as 'number'. The iteration termination condition is then determined. If the maximum number of iterations is reached, or the set optimal fitness threshold is met, the iteration ends, and the current optimal grid member position and fitness value are output. If the termination condition is not met, a candidate grid member is selected as the next grid member, the number of grid members is recalculated, and the iteration continues until the termination condition is met.
[0036] An electronic device according to the present invention, the device comprising:
[0037] Memory containing executable program code;
[0038] A processor coupled to the memory;
[0039] The processor calls the executable program code stored in the memory to execute the steps of the swarm intelligence optimization method based on grid governance as described above.
[0040] The present invention provides a computer-readable storage medium storing computer instructions, which, when invoked, are used to execute the steps of the swarm intelligence optimization method based on grid governance as described above.
[0041] Beneficial effects: Compared with the prior art, the advantages of the present invention are:
[0042] (1) The process of condensing community grid governance experience into a swarm intelligence algorithm can effectively solve the problem of finding the extreme value of complex functions;
[0043] (2) It has strong global convergence for both low-dimensional and high-dimensional continuous functions, especially for the optimization of extrema of high-dimensional functions, which has the characteristics of fewer iterations and faster convergence speed.
[0044] (3) The hyperparameter settings are simple and the tuning process is convenient. Based on the suggested default hyperparameter values, it can support the extreme value search of most performance test functions.
[0045] (4) It can be applied to the problem of multi-radar optimization deployment, simulate the collective intelligence characteristics of grid governance, form multiple grid members and members within their jurisdiction, and each member continuously approaches the position of the grid member in multiple iterations and evolutions, thereby achieving collective evolution and finally obtaining the optimal multi-radar deployment scheme, which has strong practicality. Attached Figure Description
[0046] Figure 1 This is a flowchart of the swarm intelligence optimization algorithm of the present invention;
[0047] Figure 2 It shows the two-dimensional Rosen-Brock function graph and the optimization process using the swarm intelligence optimization method based on grid governance (GGO algorithm); where (a) is the two-dimensional Rosen-Brock function graph and (b) is the optimization process using the swarm intelligence optimization method based on grid governance (GGO algorithm).
[0048] Figure 3 The graphs show the two-dimensional Table function and the optimization process using the GGO algorithm; where (a) is the two-dimensional Table function graph and (b) is the optimization process using the GGO algorithm.
[0049] Figure 4 The graphs show the two-dimensional Shubert function and the optimization process using the GGO algorithm (30-dimensional optimization); where (a) is the two-dimensional Shubert function graph and (b) is the optimization process using the GGO algorithm (30-dimensional optimization).
[0050] Figure 5 The graphs show the two-dimensional Levy function and the optimization process using the GGO algorithm (30-dimensional optimization); where (a) is the two-dimensional Levy function graph and (b) is the optimization process using the GGO algorithm (30-dimensional optimization).
[0051] Figure 6 It shows the two-dimensional Rotated Hyper-Ellipsoid function image and the optimization process using the GGO algorithm (30-dimensional optimization); where (a) is the two-dimensional Rotated Hyper-Ellipsoid function image and (b) is the optimization process using the GGO algorithm (30-dimensional optimization).
[0052] Figure 7 The graphs show the two-dimensional Sum Squares function and the optimization process using the GGO algorithm (30-dimensional optimization); where (a) is the two-dimensional Sum Squares function graph and (b) is the optimization process using the GGO algorithm (30-dimensional optimization).
[0053] Figure 8 This is a diagram illustrating a scenario of optimized deployment of multiple radars.
[0054] Figure 9 This is an image showing the optimization results of a multi-radar deployment scenario using the GGO algorithm.
[0055] Figure 10 This is a diagram illustrating the iterative process of using the GGO algorithm to solve the multi-radar optimal deployment scenario. Detailed Implementation
[0056] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby providing a clearer and more definite definition of the scope of protection of the present invention.
[0057] This invention provides a swarm intelligence optimization method based on grid governance (i.e., the Grid Governance Optimization Algorithm (GGO)), which can be used to efficiently solve the extrema and optimal solutions of complex functions. It fully draws on my country's community grid-based governance approach, decomposing complex problems into multiple grids. Each grid consists of one grid administrator and multiple grid members, with the grid administrator as the responsible entity for deploying and organizing the overall work. This invention refines the characteristics of community grid-based governance to solve problems of finding the extrema of complex functions and combinatorial optimization problems. The method includes three main processes: grid partitioning, grid evolution, and grid leadership transition. The grid partitioning process completes hyperparameter definition, grid administrator position initialization, grid administrator fitness evaluation, grid member count calculation, and grid member position generation. The grid evolution process completes community evaluation, selection of role models (optimal members), and member benchmarking learning. The grid leadership transition process completes the fitness calculation of each evolved grid member, member ranking, and selection of the next grid administrator. Through the above method, the present invention can complete the extremum optimization of low-dimensional and high-dimensional functions with simple hyperparameter settings, fewer iterations, and strong global convergence, and can be applied to combinatorial optimization problems such as multi-radar optimal deployment.
[0058] like Figure 1 As shown, a swarm intelligence optimization method based on grid governance according to the present invention includes the following steps:
[0059] (1) Mesh generation process
[0060] The optimization of the target group area involves grid division to simulate the characteristics of community grid management. This includes five steps: defining hyperparameters, initializing grid member positions, evaluating grid member fitness, calculating the number of grid members, and generating grid members.
[0061] 1) Define hyperparameters: Based on the function complexity, manually predefine each hyperparameter of the algorithm, including the number of grid cells. Grid standard member number Maximum number of iterations Offset distance between grid members and grid operators , probability of mutation of grid members Grid member learning rate .
[0062] 2) Initialize grid worker positions: Each grid corresponds to one grid worker, and the positions are initialized using uniformly distributed random numbers. The position vector of a grid worker ,in For the location variable dimension, the grid worker Represents a feasible solution to an optimization problem, each of its elements All belong to the constraint vector and Values within the range, of which , That is:
[0063]
[0064]
[0065] in, Representative generated in Within the constraint range, it follows a uniform distribution. A random number is used, and a uniformly distributed random number is used here to prevent the grid operator from being too close to the grid during initialization.
[0066] 3) Evaluate the fitness of grid workers: Calculate the fitness of each grid worker based on the form of the objective function to be solved. , Representing the The fitness value of each grid member.
[0067] 4) Calculate the number of grid members: Determine the number of grid members based on the grid member fitness. The smaller the difference between the current grid member's fitness and the optimal fitness, the more grid members should be assigned to it. The specific method for determining the number of grid members is as follows:
[0068]
[0069] in For the standard number of members in the grid, The minimum (optimal) fitness among all grid members, For maximum fitness, For the first Each grid worker's fitness value Extremely small numbers introduced to prevent division by zero. To limit the maximum number of grid members, This indicates the rounding operation.
[0070] 5) Generate mesh members: Mesh members are divided into two categories: regular members and mutated members. Regular members are generated by the mesh operator's position offset, which can be based on a preset offset distance. The calculation yields the following result:
[0071]
[0072] in, Representing the The first grid A regular member, A random number between 0 and 1, where 'a' is a randomly generated value of 0, 1, or -1. This is the preset offset distance between grid members and grid operators.
[0073] To prevent getting trapped in local optima, the concept of member mutation probability is introduced. That is, before generating each grid member, a random number is generated, and it is determined that the random number is lower than the member mutation probability value. When a variable member is generated, it is unaffected by the grid operator's position; only a random position variable satisfying the constraints is generated. If the random number is greater than the mutation probability, a regular member is generated. The method for calculating the position of the variable member is as follows:
[0074]
[0075] in, For the first The first grid One mutated member, For the location variable dimension, , These are the lower and upper bound vectors of the constraints on the position variables, respectively:
[0076] ,
[0077] (2) The evolutionary process of the grid;
[0078] The simulation of community evaluation and member learning progress includes three steps: community evaluation, selection of role models, and benchmarking learning.
[0079] 1) Community Evaluation: The fitness of each grid member is calculated, and the member with the best fitness is selected as the best member of the grid. The optimal members of each grid are ranked by fitness to obtain the globally optimal grid members. ;
[0080] 2) Selecting the best member: Select the currently globally optimal grid member. As role models, role models serve as the benchmarks for all grid members to learn from in the next step. This step can also be expanded to include various strategies for selecting role models.
[0081] 3) Benchmarking Learning: All grid members evolve to obtain new grid member positions based on the difference between their own position and the position of the benchmark (globally optimal member), specifically using the following formula for updating:
[0082]
[0083] in, Representing the evolved first The first grid Each grid member, For the first The first grid Each grid member, The preset member learning rate, A random number between 0 and 1.
[0084] That is, to generate random numbers within the range of constraints.
[0085] (3) Grid leadership transition process
[0086] The simulated community leadership election process includes three steps: fitness assessment, member ranking, and grid-based leadership transition.
[0087] 1) Fitness assessment: Calculate the fitness of each grid member after evolution;
[0088] 2) Member sorting: Sort each grid member according to their fitness value, and select the one with the best fitness and whose distance between variables in each dimension is not less than 1. of Each grid member will be considered as a candidate for the next term of grid members. Number of grid cells;
[0089] 3) Grid Reshuffle: Determine whether the iteration termination condition is met. If the maximum number of iterations is reached, or the set optimal fitness threshold is met, the iteration ends, and the current optimal grid member position and fitness value are output. If the termination condition is not met, a candidate grid member is selected as the next grid member and enters step 4 of process (1) again, continuing the iteration until the termination condition is met.
[0090] The present invention provides a swarm intelligence optimization system based on grid governance, comprising:
[0091] The grid partitioning unit is used to partition the region of the intelligent optimization population into a grid; it includes defining hyperparameters, initializing grid member positions, evaluating grid member fitness, calculating the number of grid members, and generating grid members.
[0092] The grid evolution unit is used to update the positions of grid members to achieve grid evolution; it includes: calculating the fitness of grid members within each grid and selecting the one with the best fitness as the optimal member of the grid. The optimal members of each grid are ranked by fitness to obtain the globally optimal grid members. Select the currently globally optimal grid member. As role models, all grid members evolve to obtain new grid member positions based on the difference between their own positions and the role model's positions.
[0093] The grid leadership transition unit is used for selecting new grid members. This includes: calculating the fitness of each evolved grid member in the grid evolution unit; ranking the grid members based on their fitness values; and selecting the grid member with the best fitness and whose positional variables in each dimension are not less than the offset distance between the grid member and the original grid member. of Each grid member will be considered as a candidate for the next term of grid members. The number of grid members is denoted as 'number'. The iteration termination condition is then determined. If the maximum number of iterations is reached, or the set optimal fitness threshold is met, the iteration ends, and the current optimal grid member position and fitness value are output. If the termination condition is not met, a candidate grid member is selected as the next grid member, the number of grid members is recalculated, and the iteration continues until the termination condition is met.
[0094] An electronic device according to the present invention, the device comprising:
[0095] Memory containing executable program code;
[0096] A processor coupled to the memory;
[0097] The processor calls the executable program code stored in the memory to execute the steps of the swarm intelligence optimization method based on grid governance as described above.
[0098] The present invention provides a computer-readable storage medium storing computer instructions, which, when invoked, are used to execute the steps of the swarm intelligence optimization method based on grid governance as described above.
[0099] This invention presents a swarm intelligence optimization method based on grid governance, capable of efficiently solving for the extrema and optimal solutions of complex functions. Drawing heavily on my country's community grid-based governance approach, it decomposes complex problems into multiple grids. Each grid consists of one grid administrator and multiple grid members, with the grid administrator as the responsible entity for deploying and organizing the overall work. This invention distills the characteristics of community grid-based governance to solve problems such as finding the extrema of complex functions and determining the optimal deployment scheme for multiple radars. The algorithm comprises three main processes: grid partitioning, grid evolution, and grid leadership transition.
[0100] The swarm intelligence optimization method based on grid governance described in this invention can be applied to the extreme value optimization problem of complex functions. As shown in Table 1, the algorithm is tested by using 3 sets of low-dimensional and 3 sets of high-dimensional complex functions respectively. Table 2 shows the reference range for the selection of parameters of the algorithm.
[0101] Figures 2-7 The diagram shows the iterative process of optimizing various complex functions using GGO. Table 3 shows the optimization results obtained using the GGO algorithm. Figure 2 (a) and (b) Figure 3 (a), (b), and Figure 4 As can be seen from (a) and (b), the GGO algorithm has good global convergence for low-dimensional complex functions, with a smooth iterative curve and fast convergence speed; Figure 5 (a) and (b) Figure 6 (a), (b), and Figure 7 As can be seen from (a) and (b), the GGO algorithm also has good global convergence for high-dimensional complex functions, with fewer iterations and faster convergence speed.
[0102] Table 4 shows the hyperparameter settings used during the test. It can be seen that GGO has few hyperparameters and is easy to set. Most test functions can be optimized without much adjustment of hyperparameters.
[0103] Table 1. Information on complex functions
[0104]
[0105] Table 2 Reference Range for GGO Algorithm Hyperparameter Selection
[0106]
[0107] Table 3 shows the optimization results of each performance test function using GGO.
[0108]
[0109] Table 4. Record of hyperparameter settings when optimizing each performance test function using GGO
[0110]
[0111] The swarm intelligence optimization algorithm based on grid governance described in this invention can also be applied to solving the problem of multi-radar optimal deployment.
[0112] In a multi-radar network detection system, the optimal radar deployment locations need to be planned based on the mission airspace and radar detection range. Optimal deployment of multiple radars needs to address scenarios such as... Figure 8 As shown.
[0113] With airspace coverage as the optimization objective and radar location as the optimization variable, the GGO algorithm is used to solve the optimization problem. The specific model to be optimized is as follows:
[0114]
[0115] in, This represents a feasible solution that satisfies the constraints. For the first The radar deployment information includes longitude, latitude, and altitude, with altitude considered as 0. ,Right now The first The radar deployment provides longitude, latitude, and altitude information. , The scope of the deployable area is defined by constraints. These represent the upper limits for the longitude, latitude, and altitude information of the deployable area, respectively. These are the lower limits for the longitude, latitude, and altitude information of the deployable area, respectively. Represents the radar under current deployment for the first Spatial coverage obtained after discretizing each height layer Represents the number of airspace height layers. This is the minimum airspace coverage requirement.
[0116] 1) Airspace range
[0117] See Table 5 below for airspace boundary points, with an altitude level of 20,000 meters.
[0118] Table 5 Information on airspace boundary points
[0119]
[0120] 2) Radar information
[0121] The verification was conducted using four radars with the same power parameters, as shown in Table 6 below.
[0122] Table 6 Radar Power Information
[0123]
[0124] 3) Constraints
[0125] Deployable area range: longitude 70~76°, latitude 32~36°, altitude 0;
[0126] Airspace coverage: no less than 10%;
[0127] 4) Algorithm parameters
[0128] When using GGO to solve the current radar optimization deployment problem, the hyperparameter settings are as follows:
[0129] Table 7. Record of GGO hyperparameter settings under the current radar optimization deployment problem
[0130]
[0131] 5) Solution process
[0132] The problem is solved by employing the three major processes of mesh generation, mesh evolution, and mesh replacement described in this invention, and by setting the hyperparameters as described above.
[0133] (1) During the grid division process, the range of values of the variables to be solved is divided into 10 grids. Each grid has one grid member and multiple grid members. The grid member represents the optimal solution in the current grid, that is, the optimal multi-radar deployment location.
[0134] (2) During the grid evolution process, each grid member continuously approaches the position of the grid member according to the GGO algorithm formula. During the approach process, new solution spaces are continuously explored through mutated members.
[0135] (3) During the grid transition process, select the members with the highest fitness among all grid members as the new grid members until the iteration ends, and output the current optimal solution, that is, the optimal multi-radar deployment location.
[0136] 6) Experimental Results
[0137] The GGO algorithm is used to solve the radar optimization deployment problem in the current scenario. The solution results are as follows: Figure 9 As shown, Figure 10 This is a diagram illustrating the iterative process of the GGO algorithm. Figure 9 It is evident that the radar deployment scheme obtained by the algorithm can achieve over 51% coverage of the mission airspace and satisfy constraints such as the deployment area; Figure 10 As can be seen, the algorithm finds the optimal solution in less than 20 iterations, demonstrating fast iteration speed and high optimization efficiency. The final optimal value obtained is -0.5129, and the optimal solution is shown in Table 8.
[0138] Table 8. Solution results of the radar optimization deployment problem using GGO (optimal deployment positions)
[0139]
[0140] The above description is merely an embodiment of the present invention and does not limit the scope of protection of the present invention. Any equivalent method or process transformation made based on the content of the present invention specification and drawings, or direct or indirect application in other related technical fields, shall also be included within the scope of protection of the present invention.
Claims
1. A swarm intelligence optimization method based on grid governance, characterized in that, Includes the following steps: S1. Divide the region of the intelligent optimization group into grids; This includes defining hyperparameters, initializing grid member positions, evaluating grid member fitness, calculating the number of grid members, and generating grid members; among these, evaluating grid member fitness includes calculating the fitness of each grid member based on the form of the objective function to be solved. ; The objective to be solved is the airspace coverage of radars at various altitude levels in a multi-radar network detection system. The optimization variable is the radar location. The model to be optimized is: ； in, This represents a feasible solution that satisfies the constraints. For the first Location information of radar deployment. , The first The radar deployment provides longitude, latitude, and altitude information. , The scope of the deployable area is defined by constraints. These represent the upper limits for the longitude, latitude, and altitude information of the deployable area, respectively. These are the lower limits for the longitude, latitude, and altitude information of the deployable area, respectively. Represents the radar under current deployment for the first Spatial coverage obtained after discretizing each height layer Represents the number of airspace height layers. To meet the minimum airspace coverage requirement; S2. Update the positions of grid members to achieve grid evolution; this includes: calculating the fitness of grid members within each grid and selecting the one with the best fitness as the optimal grid member. The optimal members of each grid are ranked by fitness to obtain the globally optimal grid members. Select the currently globally optimal grid member. As role models, all grid members evolve to obtain new grid member positions based on the difference between their own positions and the role model's positions. S3. Select new grid members; this includes: calculating the fitness of each grid member evolved in step S2; sorting the grid members according to their fitness values, and selecting the grid member with the best fitness and whose distance in each dimension is not less than the offset distance between the grid member and the grid manager. of Each grid member will be considered as a candidate for the next term of grid members. The number of grid members is determined. The iteration termination condition is checked. If the upper limit of the number of iterations is reached, or the set optimal fitness threshold is met, the iteration ends and the current optimal grid member position and fitness value are output. If the termination condition is not met, the candidate grid member is selected as the grid member of the next cycle and the iteration is returned to step S1 to recalculate the number of grid members. The iteration continues until the termination condition is met.
2. The swarm intelligence optimization method based on grid governance according to claim 1, characterized in that, Step S1 includes: (1) Define hyperparameters: Hyperparameters include the number of grid cells. Grid standard member number Maximum number of iterations Offset distance between grid members and grid operators , probability of mutation of grid members and grid member learning rate ; (2) Initialize grid worker positions: Each grid corresponds to one grid worker, and the grid worker is initialized using a uniformly distributed random number. The position vector of a grid worker ,in For the location variable dimension, the grid worker Represents a feasible solution to an optimization problem, each of its elements All belong to the bounded vectors under the constraints. and upper limit vector Values within the range, of which , , That is: ； ； in, Representative generated in Within the constraint range, it follows a uniform distribution. A number of random numbers, the same as the number of grid cells; (3) Calculate the number of grid members: Determine the number of grid members based on the grid member fitness. ; (4) Generating mesh members: Mesh members are divided into two categories: regular members and mutated members. Regular members are generated by the position offset of the mesh operator, based on a preset offset distance. The calculation shows that before each generation of grid members, a random number is first generated, and it is determined that the random number is lower than the member mutation probability value. When the time is right, a mutated member is generated. The mutated member is not affected by the position of the grid member. Only the position variable that meets the constraints is randomly generated. When the random number is greater than the mutation probability, a regular member is generated.
3. The swarm intelligence optimization method based on grid governance according to claim 2, characterized in that, The method for determining the number of grid members in step (3) is as follows: ； in For the standard number of members in the grid, The minimum fitness among all grid members, For maximum fitness, For the first Each grid worker's fitness value Extremely small numbers introduced to prevent division by zero. To limit the maximum number of grid members, This indicates the rounding operation.
4. The swarm intelligence optimization method based on grid governance according to claim 2, characterized in that, In step (4), regular members are generated by the grid operator's position offset, based on a preset offset distance. The calculation yields the following results: ； in, Representing the The first grid A regular member, A random number between 0 and 1, where 'a' is a randomly generated value of 0, 1, or -1. This is the preset offset distance between grid members and grid operators.
5. The swarm intelligence optimization method based on grid governance according to claim 2, characterized in that, The method for calculating the position of the mutated member in step (4) is as follows: ； in, A random number between 0 and 1 For the first The first grid A variant member.
6. The swarm intelligence optimization method based on grid governance according to claim 1, characterized in that, In step S2, all grid members evolve to obtain new grid member positions based on the difference between their own positions and the leader's position. The update formula is: ； in, Representing the evolved first The first grid Each grid member, For the first The first grid Each grid member, The preset member learning rate, A random number between 0 and 1.
7. A system for a swarm intelligence optimization method based on grid governance as described in any one of claims 1-6, characterized in that, include: Grid division unit, used to divide the area of the intelligent optimization group into grids; This includes defining hyperparameters, initializing grid member positions, evaluating grid member fitness, calculating the number of grid members, and generating grid members; The grid evolution unit is used to update the positions of grid members to achieve grid evolution; it includes: calculating the fitness of grid members within each grid and selecting the one with the best fitness as the optimal member of the grid. The optimal members of each grid are ranked by fitness to obtain the globally optimal grid members. Select the currently globally optimal grid member. As role models, all grid members evolve to obtain new grid member positions based on the difference between their own positions and the role model's positions. The grid leadership transition unit is used for selecting new grid members. This includes: calculating the fitness of each evolved grid member in the grid evolution unit; ranking the grid members based on their fitness values; and selecting the grid member with the best fitness and whose positional variables in each dimension are not less than the offset distance between the grid member and the original grid member. of Each grid member will be considered as a candidate for the next term of grid members. The number of grid members is denoted as 'number'. The iteration termination condition is then determined. If the maximum number of iterations is reached, or the set optimal fitness threshold is met, the iteration ends, and the current optimal grid member position and fitness value are output. If the termination condition is not met, a candidate grid member is selected as the next grid member, the number of grid members is recalculated, and the iteration continues until the termination condition is met.
8. An electronic device, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the steps of the swarm intelligence optimization method based on grid governance as described in any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, which, when invoked, are used to perform the steps of the swarm intelligence optimization method based on grid governance as described in any one of claims 1-6.