Study On Agent-Based Production Scheduling System

The increasing competition in market and particular demand of customer bring the unprecedented challenge to modern enterprises. In order to adapt to the new circumstances, the enterprise must take full advantage of the limite production resources to enhance their productivity. Against the background, a production scheduling system combined with the Agent technology has been established based on the MAS, and the application of Agent under the agile manufacturing mode has been researched in this article.The Unified Modeling Language (UML) has been used in this article to describe a base frame of production system. Task decomposition, task publication, task assignment and other problems in the production scheduling process have been studied systemicly. Based on analyzing agent's role and function realization method, agent has been divided and formally described. And the internal structure, the behavior method, the active flow, the communication language as well as the management pattern of the agent have been studied. Agent's conflict of information exchanging and the cooperation has been resolved by using the biding form which based on the contract net protocol. The mapping between the Agents and the function in the scheduling process is carried out, and then a multi-Agent-based production scheduling system is formed.An overall target optimization method based on the simulated annealing has been proposed in the article. This method took the time-and-cost synthetic result as the goal, a time-cost transforming objective function has been established, optimized disposition scheme was obtained in the "annealing" process, which solved the limitation problem that common resources agent and manages agent can not obtain the whole optimization.Because of the bad astringency of the simulated annealing algorithm, the genetic algorithm was introduced in order to create and filtrate thestochastic scheme; finally a superior dispatch scheme is obtained by combining genetic with simulated annealing algorithm. The study shows that the algorithm was feasible and valid.Using the JAVA programming language and taking the JADE as the platform, a multi-Agent-based production scheduling prototype system has been established and programmed combined with the JSP technology. This system contained the plan establishment, the task decomposition, task optimization assignment and other function model blocks, which was integrated with other function model blocks fulfilled by other researchers in group formed a agile manufacturing information management system.