A Novel Model For Fuzzy Multi-objective Permutation Flow Shop Scheduling Problem With Due Date And Algorithm Research

Flow shop scheduling problems(FSSP)are a class of scheduling problems emerging from modern industry,agriculture and business.It is crucial to solve FSSP effectively,in order to improve production efficiency and achieve resource conservation.In the past decade,researchers have been focus on FSSPs with fuzzy model to take into account significant uncertainty existing in real flow shop production.Among these FSSPs,the fuzzy multi-objective permutation flow shop scheduling problem(FMOPFSSP)could be widely applied to model production in reality,but its solution has not been optimized due to its NP-hard nature.Currently,a handful algorithms have been proposed to solve FSSPs using non-dominated set and decomposition based strategies.However,the development of these algorithms are still in a preliminary stage and much improvement is needed to apply them in practice.This thesis presents an investigation on an unsolved FMOPFSSP with fuzzy makespan and due date restrictions.The problem is modelled with novel fuzzy membership and comparison functions,and a novel decomposition based multi-objective scheduling algorithm is developed to achieve optimal solutions.Moreover,we present a comparison of results with different due dates,in order to improve the performance of the model in difference scenarios.Hence,we create a novel model for fuzzy multi-objective permutation flow shop scheduling problem(FMOPFSSP)in this thesis.The new model is designed to minimized the makespan and total flowtime of all jobs with fuzzy processing times and due date.Fuzzy measure of processing times and due date of each individual job is modeled as fuzzy conditions with modern fuzzy theories in this work.The processing time and due date conditions are described by trapezoidal and half trapezoidal fuzzy membership functions,respectively.The size of comparison size of fuzzy numbers is then accomplished with both centroid of fuzzy numbers and probability measure.This paper further develops a fuzzy multi-objective Pareto local search based on decomposition(FMOPLS/D)algorithm.This algorithm is constructed using decomposition-based MOPFSSP frame,and is designed to achieve multi-objective optimization with NEH heuristic method and Pareto elitism.Results of FMOPLS/D algorithm are evaluated by Hyper-Volume Indicator and Set Coverage.Results of numerical experiments are then presented to show the performance of the FMOPLS/D algorithm in fuzzy scheduling problems.We make use of the experimental results to improve local search performance with optimized set of parameters.Comparing with non-dominated sorting genetic algorithm with elitism(NSGA-?),the present FMOPLS/D algorithm could provide better solutions for optimization.Moreover,we investigate the performance of FMOPLS/D optimization under different due-date conditions.Numerical results show that the present algorithm could optimize the problem more effectively with sparser due-date distribution and relaxed limitation.In general,the novel FMOPFSSP model and FMOPLS/D algorithm presented in this thesis could effectively optimize FSSPs with significant uncertainty and biobjectives.The present model may be further used in real industry to solve complex multi-objective problems.