Modeling And Optimization Of Efficient And Energy-saving Flexible Job Shop Scheduling Problems

Flexible job shop schedulig problem is one of the most popular combinatorial optimization problems in manufacturing system,and it is of important academic value and engineering significance.Classical flexible job shop schedulig problem(FJSP)is the basis of many other scheduling problems.With the goverment paying incrasing attention to energy saving and emission reduction,the scheduling problems with considering energy consumpton objective are attracting more and more researchers.Therefore,this paper aims at studying FJSP,flexible job shop schedulig problem with contrable processing times(FJSP-CPT)and discrete flexible job shop schedulig problem(DFJSP).With regard to these three problems,we study them by condiering energy consumption,and propose both exact and approximate methods to solve them.The exact and approximate methods are respectively based on mixed integer linear programming(MILP)model and shuffled frog-leaping algorithm(SFLA).Firstly,FJSP with minimizing makespan is studied.An improved MILP model is proposed to solve small-sized problems to optimality.By analysing the differences of different neighborhood structures,a hybrid algorithm(HSFLA)that combines SFLA and a two-stage local search algorithm is proposed to solve medium-scale and large-scale instances.In HSFLA,the full active decoding method is used to reduce solution space.Finally,by solving benchmark intances,the effectiveness and superiority of both the MILP model and HSFLA are proved.Secondly,efficient and energy-saving FJSP with simultaneously minimizing makespan and total energy consumption of the workshop is studied.Two improved MILP models that consider energy consumption are proposed,and Pareto-optimal solutions of small-scale intances are solved based on the MILP model and weighting factor method.With regard to medium-scale and large-scale instances,a multi-objective hybrid SFLA(MO-HSFLA)based on Pareto dominance is proposed.In MO-HSFLA,a decoding method that considers both postponing strategy and Turning Off/On strategy is proposed based on the characteristics of energy-saving FJSP,which can reduce idle energy consumption of machine tools with keeping the makespan.Moreover,a multi-objective Tabu Search(TS)algorithm inclined to makespan and a multi-objective Variable Neighborhood Search(VNS)inclined to energy consumption are introduced to improve the local search ability of the algorithm.By solving the test instances,the effectiveness of the MILP model,decoding method and MO-HSFLA is proved.Then,with considering the contrallable processing times,efficient and energy-saving FJSP-CPT with simultaneously minimizing makespan and total energy consumption of the workshop is studied.A MILP model with considering energy consumption is firstly proposed,and Pareto-optimal solutions are obtained based on the MILP model and weighting factor method.A multi-objective hybrid SFLA(MO-HSFLA)based on Pareto dominance is proposed.In MO-HSFLA,a decoding method that considers three energy-saving strategies namely reducing processing speed,postponing strategy and Turning Off/On strategy is designed.Moreover,two multi-objective VNS algorithms that are respectively inclined to makespan and energy consumption are introduced to improve local search ability.By solving the test instances,the effectiveness of the MILP model,decoding method and MO-HSFLA is proved.Nextly,efficient and energy-saving DFJSP is studied.More specifically,DFJSP with minimizing makespan and simultaneously minimizing makespan and total energy consumption of the workshop are studied,respectively.As to makespan criterion,four MILP models based on different modeling ideas are firstly proposed to solve small-scale intances.To efficiently sovle medium-scale and large-scale instances,an efficient hybrid SFLA(HSFLA)is presented.In HSFLA,the encoding method of partial solution space is used,and a VNS is introduced to improve its local search ability.Moreover,a TS algorithm of full solution space is implemented to critical factory to expand the solution space and furter improve the local search ability of HSFLA.By solving benchmark intances,the effectiveness and superiority of four MILP models,HSFLA and local search algorithms are proved.With regard efficient and energy-saving DFJSP,a MILP model with considering energy consumption is firstly proposed to sovle small-scale intances,and Pareto-optimal solutions are obtained based on the MILP model and weighting factor method.To efficiently sovle medium-scale and large-scale instances,a MO-HSFLA based on Pareto dominance is proposed,which hybridizes two multi-objective VNS algorithms that are respectively inclined to makespan and energy consumption.By solving the test instances,the effectiveness of the MILP model and MO-HSFLA is proved.Finally,based on the above research findings,an energy management,control and scheduling prototype system is designed and developed.The system architecture and function modules are described.Moreover,a real instance of a diemaking shop is solved by the proposed algorithms,and the result shows the effectiveness and superiority of the proposed methods.In Chapter 7,the above work and innovative points of this paper are summarized.Moreover,the furture research directions are discussed.