A two-stage genetic algorithm for large size job shop scheduling problems

The majority of large size job shop scheduling problems are non-polynomial-hard (NP-hard). In the past few decades, genetic algorithms (GAs) have demonstrated considerable success in providing efficient solutions to many NP-hard optimization problems. But there is no literature available considering the optimal parameters when designing GAs. Unsuitable parameters may generate an inadequate solution for a specific scheduling problem. In this paper, we proposed a two-stage GA which attempts to firstly find the fittest control parameters, namely, number of population, probability of crossover, and probability of mutation, for a given job shop problem with a fraction of time using the optimal computing budget allocation method, and then the fittest parameters are used in the GA for a further searching operation to find the optimal solution. For large size problems, the two-stage GA can obtain optimal solutions effectively and efficiently. The method was validated based on some hard benchmark problems of job shop scheduling.