Research And Improvement Of Differential Evolution Algorithm Based On Local Search Strategy

With the development of science and technology,many problems from engineering and social fields become more and more complex,and it becomes more and more difficult to solve these problems.The traditional optimization methods have not meet the requirements for solving the current problem.As a solution to these problems,the swarm intelligence optimization algorithm can solve these problems well,so it is getting more and more attention from researchers all over the world.Differential Evolution(hereinafter referred to as DE)algorithm is a popular swarm intelligence optimization algorithm in recent years.Because of its simple operation,less control parameters and strong robustness,the algorithm has attracted increasing attention in the field of optimization.However,there are still some shortcomings in the actual optimization of DE algorithm,such as the sensitivity of the algorithm control parameters,the difficulty of selecting the appropriate mutation strategy and the insufficient of the local optimization ability of the algorithm.In view of these shortcomings,based on the research of traditional DE algorithm,two different improved DE algorithms are proposed.The main work is as follows:Firstly,the paper introduces the proposal,basic principle,operation flow and research significance of DE algorithm in detail,and gives the flow chart and pseudo-code of DE algorithm.In this paper,the research progress of DE algorithm at home and abroad is briefly expatiated,the related improved DE algorithm is classified and summarized,and the advantages and disadvantages of the algorithm are summarized.Similar to most intelligent algorithms,the DE algorithm itself lacks the local optimization ability,which leads to the slow convergence of the algorithm in the evolution process,and can not converge to the optimal solution with less evaluation times or evolutionary algebra.In order to keep the diversity of the population,prevent the algorithm from falling into the local optimal,strengthen the local search ability and speed up the convergence speed,a neighborhood search differential evolution algorithm based on Elitism learning is proposed.13 benchmark functions are used to test the proposed algorithm and compared with others state-of-the-art DE algorithms.Experiments show that the proposed algorithm has great advantages in convergence speed and accuracy.The performance of DE algorithm mainly depends on the mutation strategy and control parameters.The different control parameters often make a huge difference on the final result.Generally speaking,the control parameters and mutation strategies are pre set before solving the problem using DE algorithm.But in the process of solving the problem,it is very time-consuming to use the trial and error method to adjust the appropriate parameters.At the same time,the local search ability of the algorithm is not strong,and a small number of individuals experience stagnation in the evolution process.In view of the above shortcomings,an adaptive differential evolution algorithm based on hybrid region search strategy is proposed.The performance of the proposed algorithm is evaluated by 25 test functions.The results show that the proposed algorithm is better than several well-known algorithms in recent years.