Improving performance of evolutionary algorithm by adding elite solutions to population during evolution process——Application to deal with flexible job-shop scheduling problem
SU Zhao-feng · Computer Engineering and Applications Journal · 2010
Deficiency of population diversity always leads to premature convergence,which deeply limits performance of evolutionary algorithm.Solution quality can not be improved by simply enlarging search scale.To improve performance of evolutionary algorithm,part individuals in population are replaced with better solutions.The operation repeats for many times during evolution process according to search scale.The proposed strategy is studied on the basis of a symbiotic evolutionary algorithm which is used for dealing with complex flexible job-shop scheduling problem.Results of extensive computational simulations show that the proposed strategy shows better performance.Compared with traditional evolutionary algorithm,new proposed strategy shows higher efficiency in getting better solutions no matter whether search scale is large or not.