Scenarios based Tabu Search algorithm for robust job-shop scheduling

This paper discusses robust job-shop scheduling with uncertain processing times described by scenario approach. A kind of two-level scheduling model, which can trade off schedule optimality and risk-resisting robustness, is established. A two-stage scheduling algorithm based on TS (Tabu Search) is developed to solve the model. The two-stage algorithm matches to the two-level model. A general fast GSA(Genetic Simulated Annealing) algorithm is adopted to solve the first-level scheduling model and the solution obtained from the first level provides not only a standard performance as a threshold for a bad-scenario set but also an initial solution of a TS algorithm for the second stage. The TS algorithm of the second stage focuses on handling how to construct a neighborhood of candidate solutions for each move. The computational experiment was conducted to compare the two-stage algorithm based on TS with a two-stage algorithm based on SA (Simulated Annealing). The computational results illustrate the effectiveness and advantages of the algorithm developed in this paper.