A HYBRID GENETIC ALGORITHM APPROACH TO THE DYNAMIC SCHEDULING IN FMS

The FMS scheduling task requires scheduling a set of jobs on a finite set of resources according to the production plans to optimize some given objectives. The jobs to be scheduled together with machines and other resources are often taken as to be deterministic throughout the entire planning horizon. This paper studies the FMS real-time dynamic scheduling problem. The proposed dynamic scheduling system consists of five modules. They are dynamic database management module, hybrid genetic algorithm module, new plan download module, static database module, and knowledge base module. Dynamic database stores the data structure of the problem including the information related to parts and machines. After disturbance occurs, dynamic database can update the scheduling task data quickly according to the current production status of the system. The algorithm module is the core of the dynamic scheduling system. It directly determines the response time of the system to the disturbance. Genetic algorithm is used here to generate initial schedule as well as new ones. Static database mainly contains GA parameters and the GA approximate scheduling computation time. The knowledge base stores a great deal of heuristic rules, system status information, and the knowledge representation between them. It is used to provide suitable simulation rules for the GA algorithm.