A Predictive-reactive Approach for JSP with Uncertain Processing Times

Abstract

JSP with discretely controllable processing times (JSP-DCPT) that are perturbed in a turbulent environment is formulated, based on which, a time-cost tradeoff based predictive-reactive scheduling approach is proposed for solving the problem. In the predictive scheduling process, on the basis of a proposed three-step decomposition approach for solving JSP-DCPT, a solution initialization algorithm is presented by incorporating a hybrid algorithm of tabu search and simulated annealing and a fast elitist non-dominated sorting genetic algorithm; in the reactive scheduling process, Pareto-optimal schedules are generated, among which every schedule that is not dominated by any initial schedule can be selected as the responding schedule so as to maintain optimality of the objective that is to minimize both the makespan and the cost. Experimental simulations demonstrate the effectiveness of the proposed approach.