Research On Adaptive Differential Evolution Algorithm And Its Application

Differential Evolution Algorithm is a branch of evolutionary algorithms.It has the advantages of simple structure,few control parameters and low-time complexity.As the DE algorithm continues to evolve,it has been successfully applied in the fields of science,industry,power and transportation.However,common differential evolution algorithms face the problems of poor convergence accuracy,tendency of falling into local optima and inability of solving multiobjective optimization problem.In this paper,two enhanced DE algorithms are introduced and their efficiencies are confirmed by utilizing benchmark functions.The main contributions of this paper are as follows:(1)A stage-wise adaptive differential evolution algorithm(SADE)is proposed to overcome the issues of low convergence precision and the inclination to get stuck in local optima encountered by the standard differential evolution algorithm.Two evolutionary strategies based on different characteristics of each evolutionary period are given in the proposed algorithm,which are DE/current to pbest/1 with achieve used as a mutation strategy in the early stage of the algorithm,and DE\current to pbest\1 without achieve but with interference mechanism used as a mutation strategy in the later stage.Also,to improve the performance of the algorithm,the control parameters are also adaptively adjusted following the algorithm’s progress.It is essential to balance the global and local search capability for improving the convergence accuracy of the algorithm.This can be achieved by utilizing a parameter setting that produces high F and CR values in the early stages of the algorithm,while gradually decreases them to smaller values in the later stages.By adopting such a parameter setting,the algorithm can effectively explore the search space during prophase and then refine the solutions during anaphase.In order to enhance the efficiency and to prevent the algorithm from getting stuck in local optima,an interference mechanism is incorporated during anaphase.Comparisons were made with five other improved algorithms on 30 benchmark functions,and the findings clearly indicate that this algorithm delivers more precise results than the other alternatives.Finally,SADE is applied to the parameter estimation problem of FM acoustic waves and the results show that SADE achieves more accurate results than the other five algorithms.(2)In response to the inability of the differential evolution algorithm to effectively solve the multi-objective optimization problem,a strong Pareto-based multi-objective differential evolution algorithm is proposed,which mainly adopts the following strategies: firstly,individuals are divided into two categories according to their dominance,and different mutation strategies are adopted respectively;secondly,an adaptive control parameter is proposed to balance the convergence and diversity of the algorithm.For the learning factor F,if the current individual adaptation value is inferior to the previous individual adaptation value,then increase the value of F to increase the diversity of the algorithm,otherwise decrease the value of F to increase the convergence speed;thirdly,an improved environment selection strategy in SPEA2 is used to improve the diversity of the algorithm.The effectiveness of the algorithm is verified by combining SPEA2-DE with SPEA2 and five other multi-objective algorithms on the DTLZ and WFG benchmark functions and in five real engineering application problems.Finally,a multi-objective model for offloading tasks based on edge computing of internet of vehicles is designed,and SPEA2-DE is experimentally compared with three other algorithms in terms of energy consumption as well as time delay generated in task offloading.