The Application Of Multi-Agent Technology And Ant Algorithm In Manufacturing Grid System

With the development of the technology of computer and grid, and the constant advance of manufacturing informationization, Manufacturing Grid( MG ) emerges as the results of combining the information technology and manufacturing, to provide a general platform for Networked-Manufacturing. As a grid application in manufacturing,the MG is a distributed isomeric system, whose complexity lies in the heterogeneous and various resources and the complex actions. In order to keep different kinds of resource in the MG system, and even the whole system itself running continually and stably with high performance,This paper has applied Multi-Agent Technology and Ant Algorithm into the MG System and optimized the process of Resource Managing and Scheduling and discussed the correlative technology and theories.Firstly, this paper has introduced the structure of MG system and analysed the function of the MG Resource Scheduling Module. Based on the trait of the manufacturing industry, this paper has discussed the rules that the MG system should follow when it carry through resource scheduling, and confirmed the service quality parameters of manufacturing resources. Then it has analysed the structure and functioning mode of the MG Resource Scheduling Module, got the function scheme of the MG Resource Scheduling Module, and offered theoretic directions of introducing Multi-Agent Technology and Ant Algorithm.Based on function analysing of the structure of MG system, and introducing Multi-Agent technology,manufacturing system grid environment can be treated as distributed networked virtual collaborative manufacturing system which is made up of various manufacturing resource agent models and be used to run different kinds of manufacturing projects. In the networked environment,the agent doesn't only describe system functions,but also pays attention to phisical actions describing of low levels, so it must deal with large amount of entitative information in limited time.This needs the support of powerful calculating ability, and Grid Computing is the very effective method of networked manufacturing system real-time processing.So it set up the strong base for the application of Multi-Agent technology in MG system. Following the idea of Multi-Agenttechnology, various manufacturing resources in MG system can be treated as various encapsuled agent services,while considering the phisical configuration of manufacturing resources, we can divide one material manufacturing resource into various agents. MG system makes distributing property of manufacturing resources marked, so it needs urgently effective technological methods to solve this problem. Agent, as a sortware module, applied in MG Resource Managing System, has many advantages. It discusses detailedly these advantages in this paper.Ant Algorithm is brought forward firstly by the Italian M.Dorigo and other people, it imitates the actions when ants look for food, following the heuristic idea, by the derivational effect of the information intermedium — Pheromone, and gradually converges to the general optimum measure of the problem. So it is an effective method to solve flexibility computing problems. As Resource Scheduling problem in MG system is a well-known NP problem, and many experiments indicate that Ant Algorithm is a new effective algorithm to solve NP problems. It has strong robust property and internal distributed-collaborative property,and can combine with other algorithms easily.This paper explains the detailed trait of Ant Algorithm, and applys it in the Resource Scheduling Module in MG system to help it adapt to the strong dynamic trait of MG resource and improve the extensible trait of MG system.Finally, this paper chooses JDK1.4 as the programming environment, implements the application of Multi-Agent technology in MG Resource Managing System in software form.While it analyses the detailed aim and need of Resource Scheduling in MG system,applies Ant Algorithm in Resource Scheduling Module in MG system, carries out the optimization of Resource Scheduling,and proves the theoretical correctness and the methodological feasibility by emulational experiments.