The Research Of Solving Job-shop Scheduling Problem Based On Particle Swarm Optimization

Good production scheduling is essential to effectiveness of enterprise management and efficiency of production. Production is more of large scale and complicated with the progress in technology, and market competition is becoming stronger. Modern manufacturing is aimed to create value as much as possible through rational utilization of resources within a given time period. Job shop scheduling problem, as a classic model of scheduling problem, remains a main concern for researchers home and abroad. Thus, to realize the decision-making process in job distribution and time through proper use of optimization algorithm is considered valuable in both theory and practice. Particle swarm optimization (PSO), a representative technology in collective intelligence, characteristic of simple principle, fewer parameters and rapid convergence, has been successfully applied in a variety of industries.And it is a hot term on the frontiers of the optimization algorithm research.This paper is intended to present a deep study of how to improve the performance and applicability of standard particle swarm optimization. First it introduces the research background and progress in the job shop scheduling study. Second, by briefing the job-shop scheduling problem and outlining the current studies and improvements in particle swarm optimization, this study proposes a discrete particle swarm optimization strategy, in combination with genetic variation and simulated annealing. And then, in response to poor performance of local search, this study presents an improved discrete PSO algorithm on the basis of the third reference point. Simulation tests show that this algorithm performs better in convergence and efficiency. Also, this study has applied the improved algorithm to solve a fuzzy job shop problem with fuzzy processing time and fuzzy duedate, and has proved its feasibility through simulation tests.Finally, the study proposes adaptive discrete particle swarm optimization to converge the global extreme value, which in turn adjusts the control parameters in concert with the swarm's fitness and evolution.