Research On Multi-objective Job Shop Scheduling Problem Based On The Improved Genetic Algorithm

In the era of economic globalization, the manufacturing industry of China is facing great opportunities and intense market competitions. Companies need to meet the individual needs of customers and have the ability to rapid response the market. Scientific and effective production scheduling could enhance the ability to optimize the allocation of resources and reduce the production costs. It also can promote companies to effective organized and control the production process, improve companies’ability of schedule delivery to complete customer orders. It can make the company in the leading position of the intense global competition. Scheduling methods and related optimization techniques has become the focus of scholars.Recent years, job shop scheduling problem (JSP) has always being studied by researchers and therefore abundant theories are yielded. However, they are hard to put into practice because of most models has been constrained and simplified to single objective classical JSP, it is unrealistic assumptions.Therefore, this paper will introduce the research of JSP which combining with the actual multi-objective and flexible.Based the technical review on the domestic and foreign research of job shop scheduling and multi-objective optimization problems, this paper propose using genetic algorithm (GA) to optimize multi-objective JSP. And Combined with the actual situation of production workshop, it build a multi-objective flexible job-shop scheduling model including the Makespan criterion, the machine workload criterion and the earliness/tardiness punishment criterion. Using GA to solve the multi-objective optimization problem is prone to premature convergence and the fitness assignment is difficult. So build a new population update rules which is base on the fuzzy cluster, combination of random weighting method and preference information to improve GA. It can make the candidate solutions as evenly distributed as possible to maintain the diversity of the population, simplify the allocation of fitness and selection operations to improve the searching efficiency of the algorithm. The simulation results show that the improved GA can effectively solve the normal multi-objective flexible job shop scheduling problem. And applied the improved GA in the actual multi-objective flexible job shop scheduling is feasible. It can play a guiding role in the production practice.