Optimation Methods For Sequence Dependent Setup Time No-Wait Flowshop With Learning And Forgetting Effects

The sequence dependent setup time no-wait flowshop is one of the most important con-strained combinatorial optimization problems, which has been applied widely in industries, such as the metal, plastic, textile, chemical, food and semiconductor. Learning and forgetting effects has a significant impact on learning efficiency of production environment. The study of the problem of the sequence dependent setup time no-wait flowshop with learning and forgetting effects to minimize total flowtime and makespan is of great value both in theory and practical applications.The paper analyzes the characteristics of sequence dependent setup time no-wait flowshop problem with learning and forgetting effects and presents the new computation methods of objec-tive functions for the first time. A general position-based learning and forgetting effects model concerning job processing time is constructed. In order to speed up the computation of objec-tive value about newly generated solutions of search process, objective increment properties for job swap operators are developed. An iterated greedy heuristic to solve the considered problem is proposed, which consists of initial solution generation, local search and destruction and re-construction process. Two variable neighbourhood construction heuristics based on objective increment properties, depending on the job insertion direction, are presented, and we adopt the best one to generate the initial solution. According to the characteristics of the problem, the rule-based local search is proposed in order to avoid falling into local optimal solution. Adapted destruction and reconstruction process based on objective increment properties are implemented to intensify the diversification of search process.To verify the efficiency and effectiveness of the proposed algorithm, we adopt analysis of variance technique to analyze parameters and components of the algorithm to get the best appro-priate values of the considered problem. The proposal is compared with the best heuristics and non-deterministic algorithms for similar problems currently on standard set instances. The ex-periment results show that the proposal outperform the compared methods on both effectiveness and efficiency.