Hypothesis-test based simulated annealing for stochastic flow shop scheduling

Stochastic flow shop scheduling problem with uncertain processing times is an important optimization problem with strong engineering background, which is of the difficulties such as inaccurate estimation of optimization objective and NP-hardness. As an effective methodology in statistics, hypothesis-test (H-T) is employed and incorporated into simulated annealing (SA) to solve the stochastic flow shop scheduling. The proposed approach not only reserves the global search ability of SA, but also can reduce too much repeated search with respect to those solutions with similar performance in statistical sense to some extend. Simulation results demonstrate the feasibility and effectiveness of the proposed method by comparison with usual SA methods.