Improved Butterfly Algorithm And Its Application In Multi-mode Optimization And SVM Optimization

Butterfly optimization algorithm is a kind of swarm intelligence algorithm based on the regularity of foraging behavior between butterflies.The algorithm has simple mathematical model and strong optimization ability.It has been successfully applied to solve engineering and technical problems.However,butterfly optimization algorithm has some shortcomings,such as easy to fall into local optimum,low search accuracy and slow convergence speed.In this paper,an improved butterfly optimization algorithm based on dimension-by-dimension improvement and good point set is proposed to solve the above shortcomings of the algorithm.The good point set strategy is introduced to initialize the population uniformly,increase the diversity of the population and avoid falling into local optimum.The dimension-by-dimension improvement strategy is introduced to explore each dimension of the optimal individual in current generation.If the exploration effect of a certain dimension is good,the original value of the dimension is replaced by a new dimension value.Dimension-by-dimension improvement strategy can improve the convergence speed and accuracy of butterfly optimization algorithm.The simulation experiments on 24 benchmark functions and 6 contrast intelligent swarm optimization algorithms show that the proposed algorithm has good optimization performance in search accuracy,convergence speed and stability.A new multi-mode optimization algorithm called multi-mode butterfly optimization algorithm(MBOA)is proposed.This method combines BOA algorithm with three strategies.The first strategy involves a storage mechanism that allows efficient recording of potential local optimality based on the butterfly's fitness and distance from other potential solutions.The second strategy is designed to accelerate the detection of potential local optima.The original BOA search strategy was mainly carried out by the best individual(global best)found to date.Under the second strategy,the BOA policy is modified to be affected by the individuals contained in the storage mechanism.A third strategy is a cleaner to eliminate similar solutions that may represent the same optimal value.It has been tested on 6 benchmark functions,and the algorithm can basically find out all the global and local optimal solutions of the functions.The improved butterfly optimization algorithm is used to optimize the parameters of support vector machine.Tests were carried out on 4 UCI data sets and intrusion detection data sets KDD CUP99.The experimental results show that compared with the original algorithm and other six swarm intelligence algorithms,the proposed algorithm in this paper can select better support vector machine parameters and achieve better classification accuracy and stability.