Study On Multi-objective Dynamic Production Scheduling Problem Of Flexible Job-shop Under Uncertainty

With the development of modern manufacturing industry, the traditional job-shop scheduling methods are too simplistic to solve the actual job-shop scheduling problem. As an important extension of traditional job-shop scheduling problem, the flexible job-shop scheduling problem (FJSP) considers complex market demands and various uncertain disturbances that come out in actual production processing. In the uncertain production environment, there is theoretical and practical significance to build a multi-objective rescheduling model and to design an effective algorithm.At present, most researchs of flexible job-shop multi-objective rescheduling problem have some shortcomings. Most analysis of disturbances is based on their phenomenon, not their essential characteristic. A single rescheduling policy may make production system be unstable. The performance of multi-objective optimization algorithm could be improved. A rescheduling evaluation method is needed. To solve these problems, this thesis has made the following study.Firstly, based on the traditional FJSP model, a multi-objective rescheduling model is built, in which more optimization objectives are used. The second, the essential characteristic of disturbances is analyzed. The concept of shift coefficient is put forward to reflect the influences of disturbances. And a hybrid-driven rescheduling policy based on shift coefficient is put forward, responding to various disturbances. The third, based on the different types of disturbances, the detailed rescheduling operations are put forward. The fourth, by analyzing the deficiency of traditional Non-dominated Sorting Genetic Algorithm-II (NSGA-II), three algorithm optimization strategies are put forward:mutation strategy based on evolution degree, mutation strategy based on blood relative, bottom disruption strategy. Rescheduling scheme could be got by the modified NSGA-II.At last, penalty cost including transportation cost and tardiness cost is put forward. And it gets rid of the non-optimal rescheduling schemes in the process of multi-objective optimization by considering penalty cost. By comparing with the original scheme, the performance of the last rescheduling scheme can be evaluated, then whether the rescheduling scheme should be adopted can be decided.