Research On Improvement And Application Of Differential Evolution Algorithm

Differential evolution algorithm is a global optimization algorithm based on population.The algorithm mainly adopts difference mutation operator,crossover operator and selection operator to make the population iterate continuously and finally approach or get the global optimal solution.The algorithm is simple and easy to implement,fast convergence speed,fewer parameters and so on.Because of its great application potential and development prospect,it has attracted extensive attention from researchers at home and abroad.In this thesis,based on the existing research results of differential evolution algorithm,the double-population co-evolution differential evolution algorithm and the constrained differential evolution algorithm with ensemble of mutation strategies are proposed.The main research contents are as follows:1.The double-population co-evolution differential evolution algorithm is proposed.Mainly combining the “DE/current-to-pbest/1” with archive and “DE/current-to-rand/1” mutation strategies.At the same time,the population division technology is introduced,and the parameters of “DE/current-to-pbest/1” are modified by adaptive method.In order to verify the performance of the proposed algorithm,numerical experiments were carried out for30 unconstrained optimization test problems.The results show that the double-population co-evolution differential evolution algorithm has certain advantages compared with other algorithms.2.The constrained differential evolution algorithm with ensemble of mutation strategies is proposed.In order to deal with the problem of constrained optimization effectively,the relation between constrained condition and the objective function is balanced by the method of ensemble of mutation strategies difference evolution and feasible rule constrained method.The numerical experiment results of 24 constrained optimization test problems show that the constrained differential evolution algorithm with ensemble of mutation strategies has certain competitiveness.3.The two improved differential evolution algorithms are respectively applied to solve two kinds of practical problems,namely,the traveling salesman problem and the logistics center location problem.The simulation results show the effectiveness of the improved differential evolution algorithm.