A Hybrid of Variable Neighbor Search and Fuzzy Logic for the permutation flowshop scheduling problem with predictive maintenance

This study focuses on permutation flowshop scheduling problem (PFSP) under availability constraints with makespan and maintenance cost optimization criteria. Machines unavailabilities are due to predictive maintenance interventions scheduled based on Prognostics and Health Management (PHM) results. Hence, we deal with the post prognostic decision making in order to improve system safety and avoid downtime and inopportune maintenance spending. For this reason, we propose a new interpretation of PHM outputs to define machines degradations corresponding to each job. Moreover, to take into account the several sources of uncertainty in the prognosis process, we choose to model PHM outputs using fuzzy logic. Motivated by the computational complexity of the problem, Variable Neighborhood Search (VNS) methods are developed including well designed local search procedures. Computational experiments carried out on well known benchmark sets for permutation flowshop show that the proposed algorithms seems to be efficient and effective.