Stochastic mixed-model assembly line sequencing problem: Mathematical modeling and Q-learning based simulated annealing hyper-heuristics

This paper presents a mixed-model sequencing problem with stochastic processing times (MMSPSP) in a multi-station assembly line. A new mixed-integer nonlinear programing model is developed to minimize weighted sum of expected total work-overload and idleness, which is converted into a mixed-integer linear programming model to deal with small-sized instances optimally. Due to the NP-hardness of the problem, this paper develops a novel hyper simulated annealing (HSA). The HSA employs a Q-learning algorithm to select appropriate heuristics through its search process. Numerical results are presented on several test instances and benchmark problems from the related literature. The results of statistical analysis indicate that the HSA is quite competitive in comparison with optimization software packages, and is significantly superior to several SA-based algorithms. The results highlight the advantages of the MMSPSP in comparison with traditional deterministic approaches in mixed-model sequencing contexts.