Resilience-Based Surrogate Robustness Measure and Optimization Method for Robust Job-Shop Scheduling

Abstract

This paper addresses the robust job-shop scheduling problems (RJSSP) with stochastic deteriorating processing times by considering the resilience of the production schedule. To deal with the disturbances caused by the processing time variations, the expected deviation between the realized makespan and the initial makespan is adopted to measure the robustness of a schedule. A surrogate model for robust scheduling is proposed, which can optimize both the schedule performance and robustness of RJSSP. Specifically, the computational burden of simulation is considered a deficiency for robustness evaluation under the disturbance of stochastic processing times. Therefore, a resilience-based surrogate robustness measure (SRM-R) is provided for the robustness estimation in the surrogate model. The proposed SRM-R considers the production resilience and can utilize the available information on stochastic deteriorating processing times and slack times in the schedule structure by analyzing the disturbance propagation of the correlated operations in the schedule. Finally, a multi-objective hybrid estimation of distribution algorithm is employed to obtain the Pareto optimal solutions of RJSSP. The simulation experiment results show that the presented SRM-R is effective and can provide the Pareto solutions with a lower computational burden. Furthermore, an RJSSP case derived from the manufacturing environment demonstrates that the proposed approach can generate satisfactory robust solutions with significantly improved computational efficiency.