Research On Distributed Flexible Job-shop Scheduling Problem Based On Improved Artificial Bee Colony Algorithm

As an important pillar of China’s national economy,with the deepening of economic globalization and the acceleration of economic integration in recent years,manufacturing has shifted from traditional small-scale production to large-scale manufacturing.With the expansion of enterprises and the establishment of transnational enterprises,enterprises deploy production and manufacturing bases all over the country and even the world,and distributed manufacturing has developed into one of the main manufacturing modes at present.As an important distributed manufacturing mode,distributed flexible manufacturing is favored more and more by enterprises because of flexible factories distributed in different places.It is because of the flexibility of distributed flexible job shop that the reasonable allocation of processing tasks and the scheduling of flexible factories are more important to the production efficiency of enterprises.Therefore,in order for enterprises to better schedule the distributed flexible manufacturing process,this paper take the artificial bee colony algorithm(ABC)as the method to solve the distributed flexible job shop scheduling problem(DFJSP)under various optimization objective conditions,through establishing mathematical model of the problem and improving the characteristics of the algorithm.The main contents of this paper are as follows.(1)For the distributed flexible Job-shop scheduling problem with makespan as the optimization objective,an improved genetic artificial bee colony algorithm(GABC)was proposed.The specific improvement includes discrete improvement of artificial bee colony algorithm combined with genetic algorithm(GA).A coding scheme based on machine coding was proposed.According to the coding characteristics and the characteristics of distributed flexible job shop,a crossover operation based on coding similarity was designed to avoid illegal solutions generated by crossover and improve the efficiency of the algorithm.The search operation in the employed bee phase was improved,after crossover operation,two mutation operations were carried out with different probabilities.The nectar source discarding mechanism of scouts bee phase was improved,and the nectar source reaching the upper limit was partially discarded by comparing with the global optimal solution,so as to prevent the damaged high-quality solution from falling into the random search again.Finally,different algorithms were used to compare the solutions of examples to verify the efficiency of the improved genetic bee colony algorithm.(2)An artificial bee colony algorithm based on empirical strategy(ESABC)was proposed to solve the distributed flexible job-shop scheduling problem with makespan and critical machine loads as optimization objectives.The algorithm was improved as follows,the framework of the algorithm is improved,an experienced bee mechanism is proposed,which was used to record the same or better nectar sources of each colony in the process of searching nectar sources,providing guidance for the colony search.An adaptive crossover mechanism of coding similarity was used to balance the global search ability of different nectar sources and different operating stages of the algorithm.In the onlooker bee phase,a variety of neighborhood transformation methods based on optimization objectives were designed to improve the local search efficiency.The method of generating nectar source was improved by using experience bees to search the neighborhood.Finally,the improved algorithm was proved to be superior in convergence and solution performance by comparing the algorithm with computational experiments.(3)A green distributed flexible job-shop scheduling model with transport constraints was proposed.The model considers the transportation cost between the warehouse and the remote factory,the transportation cost between the machines in the factory,and the energy consumption in the processing and transportation process,and carried out multi-objective optimization with makespan,total transport cost and total energy consumption.A multi-population artificial bee colony algorithm(MPABC)was proposed to solve the problem.The algorithm designed a new population division mechanism,and adopted different search methods for different sub-employed bee colonies,and improved the sub-onlooker bee colony following mechanism.Finally,the effectiveness of the improved algorithm was verified by computational experiments of different algorithms.