Research On Strategy Optimization For Brain Storm Optimization Algorithm

The brain storm optimization algorithm has a simple structure and is easy to understand.The core of the algorithm is to simulate humans using brainstorming meetings to solve problems.It is the only swarm intelligence optimization algorithm inspired by human group behavior,and has strong global optimization ability and algorithm robustness.The brain storm optimization algorithm uses clustering strategy to divide the initial population into several smaller search spaces,which is convenient for the algorithm to perform local search and achieve algorithm convergence.When a new individual is generated,the BSO algorithm uses probability parameters to control whether the new individual is generated based on one class or two classes.At the same time,in order to achieve algorithm divergence and increase the population diversity,the algorithm uses mutation strategy to add noise to the new individual.The brain storm optimization algorithm incorporates convergence and divergence operations in each iteration,which reflects the algorithm's strong adaptability and structural scalability.Therefore,different classification and mutation strategies will have different effects on the algorithm.This paper analyzes the probability parameter setting,classification strategy and mutation strategy in the brain storm optimization algorithm,and makes some improvements to the original brain storm optimization algorithm.The main results are as follows:1.Through the experimental analysis of the probability parameter clusterp,this paper sets it as a dynamic parameter that increases with the number of iterations,speeds up the convergence speed in the early stage of algorithm optimization,and strengthens the global optimization ability in the later stage.This paper proposes a new algorithm called FIBSO by analyzing the advantages and disadvantages of the k-means clustering strategy in the original brainstorming optimization algorithm.The FIBSO algorithm uses classification based on fitness values to replace the k-means clustering in the original algorithm,which speeds up the convergence speed of the algorithm and increases the population diversity of the algorithm.At the same time,the FIBSO algorithm uses dynamic probability parameters to increase the probability of generating new individuals from the two classes in the early stage of algorithm optimization and speed up the convergence speed.In the later stage of algorithm optimization,the FIBSO algorithm uses dynamic parameters to increase the probability of generating new individuals from a class and enhance the ability of the algorithm to locally optimize.The experiment uses 4 classic standard test functions to test the FIBSO algorithm,and compares the results with the BSO,PSO,and GA algorithms to prove the effectiveness of the improved strategy.2.In the original brain storming optimization algorithm,Gaussian mutation was used to generate new individuals.Gaussian mutation is a common mutation method,and the amount of mutation produced is limited,so it is easier to fall into local extremes when the brain storm optimization algorithm is optimizing.Aiming at this problem,a brain storm optimization algorithm based on multi-branch chaotic mutation is proposed.When the original BSO algorithm falls into a local optimum,the MCMBSO algorithm uses multi-branch chaotic mutation instead of Gaussian mutation in the original algorithm to generate new individuals.The MCMBSO algorithm uses a variety of chaotic sequences to expand the variation range of new individuals and enhance the algorithm's global optimization capability.The experiment uses 10 classic test functions with different dimensional problems for testing,and compares the experimental results with BSO,PSO,GA,and CS algorithms.The experimental results show that the proposed algorithm MCMBSO can effectively avoid falling into local optimum,and has higher stability and global search ability.