Minimizing makespan for scheduling stochastic job shop with random breakdown

This paper addresses the problem of scheduling stochastic job shop subject to breakdown. A relative good and efficient genetic algorithm (GA) is proposed for the problem with normal processing time, resumable jobs and the objective of minimizing makespan. Some operations of normal processing times are defined to build the schedule. In the GA, an operation-based representation is used, a discrete event driven decoding method is presented to deal with breakdown and repair, and generalized order crossover and swap are applied to produce new solutions. Genetic operators are separate from the handling of random breakdown. The GA is applied to some test problems and compared with a simulated annealing (SA) and a particle swarm optimization (PSO). The computational results show the GA performs better than PSO and SA for stochastic job shop scheduling problems considered.