Application on job-shop scheduling with Genetic Algorithm based on the mixed strategy

Abstract

Adaptive genetic algorithm for solving job-shop scheduling problems has the defects of the slow convergence speed on the early stage and it is easy to trap into local optimal solutions, this paper introduces a time operator depending on the time evolution to solve this problem. Its purpose is to overcome the defect of adaptive genetic algorithm whose crossover and mutation probability can not make a corresponding adjustment with evolutionary process. Algorithm's structure is hierarchical, scheduling problems can be fully demonstrated the characteristics by using this strategy, not only improve the convergence rate but also maintain the diversity of the population, furthermore avoid premature. The population in the same layer evolve with two goals-time optimal and cost optimal at the same time, the basic genetic algorithm is applied between layers. The improved algorithm was tested by Muth and Thompson benchmarks, the results show that the optimized algorithm is highly efficient and improves both the quality of solutions and speed of convergence.