Multi-objective evolutionary simulation based optimization mechanism for a novel stochastic reliability centered maintenance problem

Research paper by Seyed Habib A. Rahmati, Abbas Ahmadi; Behrooz Karimi

Indexed on: 03 May '18Published on: 15 Apr '18Published in: Swarm and Evolutionary Computation


Publication date: Available online 23 March 2018 Source:Swarm and Evolutionary Computation Author(s): Seyed Habib A. Rahmati, Abbas Ahmadi, Behrooz Karimi This research develops a novel stochastic reliability-centered maintenance (RCM) mechanism within a new multi-objective joint maintenance and production planning problem. RCM in this integrated problem is an agent that monitors and manages the maintenance functions of a stochastic complex production-planning problem, namely flexible job shop scheduling problem (FJSP). The novel developed RCM takes benefit of stochastic condition based maintenance (CBM) approach that works based on stochastic shocking scheme of machines during their process time. It activates the maintenance activities, including preventive and corrective maintenance, according to the degradation level of system reliability after shocks and not merely according to the predetermined intervals. In addition to the maintenance activation times, the maintenance durations of different kinds are also modeled stochastically. Furthermore, different types of stochastic maintenance costs are also considered alongside system reliability and complementation time (C max). Moreover, as the problem belongs to the NP-Hard class of optimization problems, four multi-objective simulation based optimization (SBO) algorithms, called multi-objective biogeography based optimization (MOBBO) algorithm, Pareto envelope-based selection algorithm (PESA), new version of non-dominated sorting genetic algorithm (NSGAIII) and multi-objective evolutionary algorithm based on decomposition (MOEAD) are employed to solve the underlying problem. A novel visualization approach joint by Gant chart is also proposed to discuss the whole RCM scheme, systematically. Different test problems, statistical tests and outputs explain the problem and algorithms' performance explicitly.