Evolutionary algorithms for multi-objective flexible job shop cell scheduling

Abstract The multi-objective flexible job shop scheduling in a cellular manufacturing environment is a challenging real-world problem. This recently introduced scheduling problem variant considers exceptional parts, intercellular moves, intercellular transportation times, sequence-dependent family setup times, and recirculation requiring minimization of makespan and total tardiness, simultaneously. A previous study shows that the exact solver based on mixed-integer nonlinear programming model fails to find an optimal solution to each of the ‘medium’ to ‘large’ size instances considering even the simplified version of the problem. In this study, we present evolutionary algorithms for solving that bi-objective problem and apply genetic and memetic algorithms that use three different scalarization methods, including weighted sum, conic, and tchebycheff. The performance of all evolutionary algorithms with various configurations is investigated across forty-three benchmark instances from ‘small’ to ‘large’ size including a large real-world problem instance. The experimental results show that the transgenerational memetic algorithm using weighted sum outperforms the rest producing the best-known results for almost all bi-objective flexible job shop cell scheduling instances, in overall.