The Improvement Of Two Typical Intelligent Optimization Algorithms

In engineering and technology fields,optimization problems are often encoun-tered,such as the minimum parameter error,the optimal allocation and the invest-ment decision and so on.Although there are many traditional methods for optimiza-tion problems,they had strict requirements for the analyticity of their objectives and couldn't meet the needs of solving practical problems.With the development of biology science,various intelligence algorithms have been proposed by researchers and provided new approaches for optimization problems by simulating the popula-tion behavior of natural species.Specially,as the typical algorithms for optimization problems,differential evolution algorithm and artificial bee colony algorithm have attracted widespread attentions from scholars,because of their simple principle,few parameters and easy implementation.However,they have some drawbacks in the search process,such as slow convergence,worse accuracy and so on.To overcome these shortcomings,this thesis improve algorithms respectively.The main contribu-tions are follows:1.To alleviate the drawbacks of CoDE that the chosen strategies axe weak exploitative and the initial population always have relative poor performance.First,the opposite learning is utilized to improve the performance of the initial solution.Next,a new mutation strategy is designed by using elite solution information and added in the mutation strategy pool to enhance the exploitation of algorithm.2.To further enhance the search ability of artificial bee colony algorithm and maintain the diversity of population,an improved bee colony algorithm with greedy search is proposed.First,in the employed bees phase,an update strategy is de-signed based on the idea of greedy search.This strategy not only has strong global exploration,but also improves exploitation.Next,in the onlooker bees phase,a combinatorial strategy with greedy search is employed to balance the exploitation on food sources and the diversity of bee colony.The proposed algorithm can improves the optimization accuracy effectively.Comparison with the existing intelligent algorithms,the numerical results on 30 benchmark functions from CEC2014 show that the proposed algorithms have faster convergence and better optimization accuracy.