New Evolutionary Subset: Application to Symbiotic Evolutionary Algorithm for Job-Shop Scheduling Problem

Evolutionary algorithms (EAs) prove to be powerful in solving combinatorial optimization problems. Construction mechanism of evolutionary subset affects the search capacity and efficiency. In this paper, random evolutionary subset is proposed to promote EAs´ performance and compared with traditional neighborhood evolutionary subset. During the evolution process, evolutionary subsets are formed for localized evolution. To construct neighborhood evolutionary subset all individuals are chosen from a neighborhood in the population while in random evolutionary subset randomly. Two parameters affect the performance of neighborhood evolutionary subset: size and location. For random evolutionary subset, size is the only parameter that affects the evolution process and the suitable value is 6-15 on the basis of experimental results. The experimental results show the random evolutionary subset has better performance: showing high efficiency in getting much better solutions and fairly well solutions are obtained in early stage of evolution process.