Study And Application Of Dynamic Scheduling For Multi-Objective And Flexible Manufacturing In MES

Manufacturing Execution Systems (MES) is a management information system oriented to shopfloor. It plans to fill the gap between ERP layer and shop floor control system layer. Reasonable and optimized job level production scheduling plays a key role in successfully implementing MES in manufacturing enterprises. With technologies advance, flexible, dynamic, and multi-objective is the direction of production scheduling problem research field which has a long history.The application background of this paper is manufacturing execution system of a discrete manufacturing enterprise. This paper studies flexible manufacturing dynamic multi-objective scheduling which is suited to current production practice and in line with the development direction of scheduling theory. After fully analyzing the actual production of flexible manufacturing environment nowadays, based on the FJSP (flexible job-shop scheduling problem) model, a multi-objective and flexible job-shop scheduling dynamic model is constructed, which is being addressed in this paper.To solve the dynamic multi-objective FMS scheduling problem, an event-driven approach is adopted to deal with dynamic disturbance from in- and/or out-side of the shop floor; and use a rolling horizon mechanism to reduce the scale of the problem. In generating the optimized scheduling plan, Genetic Algorithm (GA) is being used, which has been wildly used in production scheduling theory and multi-objective optimization research filed. However, standard GA is insufficient to solve the optimization problems which have flexible, dynamic, multi-objective features. So an improved GA is presented in this paper, specific innovation and improvement as follows: a new encoding method for FJSP; domain objects particpate in genetic manipulation, thereby self-conduct mutation becomes possible; Ideal Trival Solution (ITS) is proposed from Pareto optimal, which has widespread significance for multio-objective optimization filed; and based on ITS, the individuals can be evaluated against multi-metrics; in order to speed up GA, explicit parallel is enabled for evolving process.Finally, on the basis of the above, based on modular, reconfigurable, easy integration of real-time dynamic scheduling system framework, prototype system is designed and implemented for simulation test; and the quality of the product is verified by application examples results in testing.