The Study On Multi-objective FJSP Based On Improved NSGA-Ⅱ

Job-shop Scheduling is an important foundation of manufacturing system, and its core is optimization technology. Near these years, customers’personal demands tc product is more and more.Traditional large quantity and single specie productior method is difficult to adapt to changing market conditions, and it is insteaded by small quantity and multi-species production method. It always needs to optimize multiple targets in actual production scheduling process. However, the dimensions of the multiple targets are always differernt. Also, it is always conflicting between multiple targets. When one target is optimize, other targets might become deteriorational.Under this background, the multi-objective flexible job shop scheduling problem (FJSP) is considered as study objective. In the direction of production scheduling theory and multi-objective optimization methods theory, the modeling and algorithm of multi-objective FJSP is focusly researched. Considering the characteristic of modern manufacturing system, the multi-objective FJSP model which includes six scheduling indexes:makespan, total machine load, the bottleneck machine load, average flow through time, delivery and production costs is built. To the deficiency of NSGA-II algorithm in solving multi-objective optimization problem, an improved NSGA-II algorithm is advanced.Based on original NSGA-II algorithm, a self-adaptive strategy which can dynamically adjust crossover and mutation probability with the changing of evolution algebra is introduced, which can improve the global and local optimization ability of evolution algorithm, so as to aviod algorithm precocity. At the same time, it can prevent algorithm prevent falling into local optima. A selection strategy of elite individuals is put forward, which can improve the diversity of the population. Another, an improved elitist base on exclusion mechanism and circular crowding distance sorting. Exclude unreasonable solutions according a setted smaller value L before non-dominated sorting at first, and then calculate crowding distance on every non-dominated cutting edge cycly. At last, select individuals with lagrer crowding distance on every non-dominated cutting edge as next generation population.Finally, the improved NSGA-II algorithm is implemented by C++ programming language. The improved NSGA-II algorithm is used in benchmark of FJSP with eight workpieces and eight mechines. By comparing the results of improved NSGA-II with other multi-objective evolutionary algorithm to test the effectiveness of improved algorithm in solving multi-objective FJSR.And then, the improved NSGA-II algorithm is used in solving real production activities of manufacturing plant. It is proved that the improved NSGA-II algorithm can get satisfactory Pareto solutions and scheduling schemes. Finally, Analytic Hierarchy Process is used to evaluate Pareto solutions solved by improved NSGA-II algorithm.