Scatter Search For Distributed Assembly Flowshop Scheduling To Minimize Total Tardiness

The distributed assembly permutation flowshop problem (DAPFSP) is a typical NP-hard combinatorial optimization problem and represents an important area in multiple distributed production systems, where a series of jobs are to be processed on machines in one of specific factories (processing stage) and then assembled into final products by a single machine (assembly stage). In this study, a class of complex DAPFSP with respect to minimizing the total tardiness is investigated by considering real-practice constraints, including no-wait, no-idle and due dates, labeled as DAPFSP-T hereafter. And an effective scatter search based memetic algorithm (SS-MA) is proposed to address the difficulties of proposed DAPFSP-T model. Specifically, in the proposed SS-MA, a few compositive heuristics are proposed by comprehensively deploying several well-known constructive heuristics, and serve as the initialization method. Subset generation mechanism and solution combination methods are deliberatively designed to implement the global coarse exploration. Meanwhile, the improvement procedure in conventional SS is incarnated by a three-stage simulated annealing (3SSA) with three kinds of neighborhood structures to perform the local fine exploitation. Through the above sophisticated combination of multiple operators, it is expected in the proposed SS-MA the balance between global and local search abilities could be well achieved. And it is demonstrated by our experimental studies and comparisons that the proposed SS-MA could yield satisfactory searching performances, where solutions are significantly improved compared with multiple heuristic methods.