A Knowledge-Driven Cooperative Estimation of Distribution Algorithm for Energy-Efficient Reconfigurable Distributed Flowshop Group Scheduling Problem

Abstract

With the growing energy problems and globalization, energy-efficient scheduling and distributed manufacturing are receiving more attention from researchers and manufacturers. According to the processing printed circuit boards in real-life cellular manufacturing systems, we consider an energy-efficient reconfigurable distributed flowshop group scheduling problem (EE_RDFGSP) with the minimization of makespan and total energy consumption. This paper establishes a mixed-integer linear programming model to tackle the proposed problem. Since the multiple coupled subproblems between two flows of EE_RDFGSP, a knowledge-driven cooperative estimation of distribution algorithm is presented. First, multilayer cooperative probability models are designed for capturing multi-dimensional valuable information from superior solutions. Second, a knowledge-driven decoding scheme with an energy-saving strategy is provided to refine the obtained solutions, which is based on the probability models. Third, a hybrid initialization approach is designed to offer an initial population, and an effective sampling strategy for the probability models is devised to generate individuals with the extracted knowledge in the probability models. Moreover, local intensification with knowledge-based operators is employed to improve the exploration. Finally, the statistical results illustrate the effectiveness of the proposed algorithm for addressing the EE_RDFGSP.

Keywords: Distributed flowshop scheduling, reconfigurability, group scheduling, estimation of distribution algorithm, energy-efficient scheduling