Research On Green Scheduling Problem Of Discrete Industrial Process Based On Imperialist Competitive Algorithm

Manufacturing industry is not only an important pillar of our national economy but also the major source of energy and environmental problems.With the rapid growth of national economy and the continuous deterioration of environment,green manufacturing,which is a modern manufacturing mode,has attracted extensive attention of academia and industry.As an important part of manufacturing industry,discrete industry urgently needs to improve economic indicators and reduce the impact on environment.Green scheduling of discrete industrial process can achieve energy saving,emission reduction and consumption reduction,economic benefit improvement and realize green manufacturing process at the same time through the reasonable optimization of resource allocation,operation sequence and execution pattern,which has higher practical significance and application value.This dissertation focuses on the green scheduling problems of discrete industrial process,which mainly includes green flexible job shop scheduling problem(FJSP)and green hybrid flow shop scheduling problem(HFSP),and deeply studies the green FJSP with the same degree of objective importance and different constraints and the above two kinds of green scheduling problems with different degree of objective importance and considering key objectives,and combines new optimization mechanisms such as feedback and empire level to design a variety of novel empire competitive algorithms to obtain high-quality optimization results.The main research works of this dissertation are as follows.(1)Green FJSP with total energy consumption constraint is considered and a twophase meta-heuristic algorithm based on imperialist competitive algorithm(ICA)and variable neighborhood search(VNS)is proposed.The goal is to minimize makespan and total tardiness under the condition that total energy consumption does not exceed a given threshold.In first phase,the original problem is converted to the green FJSP with three objectives including total energy consumption,and a novel ICA is designed to solve above problem.An energy consumption threshold is obtained based on the optimization results of the novel ICA.In second phase,an efficient VNS is proposed to solve the original problem by using new methods for comparing solutions and updating the non-dominated set.The computational results show that the two-phase algorithm has strong advantages for the considered green FJSP.(2)A green FJSP with transportation is investigated and an ICA with feedback is developed to minimize makespan,total tardiness and total energy consumption simultaneously,in which the paremeters of assimilation and revolution and adaptive selection neighborhood structure of are determinted by feedback,not all colonies perform assimilation in an empire,each selected colony is assimilated more than once and has more than one learning object.A new imperialist competition is presented to enhance information communication between empires and reinforce the search of some worst solutions in population.Extensive experiments validate that the ICA with feedback has superiorities on solving the considered problem.(3)An improved ICA is proposed to solve many-objective green FJSP with the simultaneous minimization of total energy consumption,maximum tardiness,makespan and maximum workload of machine.Initial empires are newly constructed so that most of imperialists are assigned close number of colonies.Assimilation of imperialist is introduced,and revolution and imperialist competition are implemented in a new way.Extensive experimental results validate that the significant impact of strategies of the improved ICA on its performance and the proposed algorithm has strong advantages on solving many-objective green FJSP.(4)Green FJSP with key objectives is investigated and a diversified ICA is proposed to optimize fully makespan and total tardiness as key objectives and continuously improve total energy consumption as non-key one.In the diversified ICA,a new assimilation is implemented by making each solution has more than one learning object and differing some best solutions from other colonies and a new imperialist competition is done by the new definition of normalized total cost and the inclusion of global search of imperialist.A number of experiments are conducted to analyze the relation between the deterioration degree of total energy consumption and improvement degree of the key objectives.Experimental results verify the effectiveness and advantages of the proposed ICA on solving the considered FJSP.(5)A green HFSP with total tardiness,makespan and total energy consumption is addressed,in which the third objective has lower importance than other ones.A new dominance relationship is defined to deal with the relative importance of objectives and a two-level ICA is presented,in which the first level consists of the strongest empire and the second one is composed of other empires.In two-level ICA,assimilation and revolution are executed differently in empires in the different search stages,the strongest empire is excluded from imperialist competition.Memory is combined with the strongest empire to construct now empire and a randomly selected member of memory is added into the winning empire to avoid the inclusion of the weakest colony of the weakest empire.Extensive experiments are conducted and the computational results show that the two-level ICA provides stronger search advantages for the above green HFSP.