Research On Improvements And Applications Of Butterfly Optimization Algorithm

The Butterfly Optimization Algorithm(BOA)is an intelligent optimization algorithm proposed in 2019,which simulates the foraging behavior of butterfly population.This algorithm has the advantages of few parameters,easy-to-understand and strong searching ability,so it is widely studied and applied.With the in-depth study of BOA,scholars found that the performance of BOA is deeply affected by poor population diversity,low optimization accuracy,slow convergence rate and other shortcomings,which to some extent limited the theoretical development and application of butterfly optimization algorithm.In order to improve the performance of BOA and expand the application range of the algorithm,this paper makes a series of improvements based on the existing problems of BOA,and studies the application of the improved BOA in function optimization,feature selection and engineering structure optimization.At the same time,the performance of the improved algorithm is further tested.The main research work of this paper is as follows:(1)To improve BOA’s performance,a self-adaptive Gaussian keyhole imaging butterfly optimization algorithm based on Latin Hypercube sampling is proposed hypercube sampling,(SGLBOA).Firstly,the initializing strategy of Latin hypercube sampling population is used to improve the diversity of population,so as to improve the possibility of finding the optimal value.Then,an adaptive optimal guidance strategy which automatically adjusts the search range in different evolutionary periods is introduced to balance the global and local search capabilities of the algorithm,so as to improve the optimization accuracy of the algorithm.Finally,Gaussian keyhole imaging strategy is used to disturb the optimal individuals,so that the population individuals are close to the optimal individuals,so as to further improve the optimization accuracy of the algorithm and accelerate the convergence rate of the algorithm.The simulation results show that SGLBOA has better convergence accuracy,convergence speed,robustness and expansibility in classical test functions,and also verifies that SGLBOA has certain advantages in solving high-dimensional classical function optimization problems.(2)In order to further improve the convergence accuracy,convergence speed and algorithm robustness of BOA on complex functions,in this paper,a dynamic adaptive differential evolution butterfly optimization algorithm based on inverse S inertia weights is proposed inverted S inertia weights,(SHDBOA).Firstly,the traditional global search strategy of BOA is replaced by an adaptive parameter differential evolution algorithm which is based on historical success and has stronger global search ability.Secondly,in the local search stage of butterfly optimization algorithm,an adaptive optimal guidance strategy based on inverted Stype inertia weight is introduced to automatically adjust the search range,so as to further improve the optimization performance of the algorithm.Finally,the dynamic adaptive information sharing strategy is introduced to improve the information exchange between population individuals in the search stage.The simulation results show that SHDBOA is superior to other comparison algorithms in terms of convergence speed,solving accuracy and solving stability,both for classical test functions and complex CEC2017 test functions.(3)In order to further verify the performance of the improved butterfly optimization algorithm in practical applications,the application of SGLBOA and SHDBOA in feature selection problems and engineering structure optimization problems is studied.The simulation results show that compared with other algorithms,SHDBOA has higher average classification accuracy in feature selection,and the Latin Hypercube population initialization strategy in SGLBOA has certain advantages in the average number of features.At the same time,the engineering structure optimization problem further verifies that compared with SGLBOA,the improved SHDBOA has more advantages in solving the engineering structure optimization problem.