| Along with the developing of the distributed control of the modern industry,people have more demands for the total performance of the control systems. So we need to explore a kind of control system,which is not just composed of one model,but of many subsystems.The paper points out the two develop direction of the distributed control system under the foundation of introducing the development of Agent,MAS system and distributed control systems,one of them is intelligentizing the distributed control subsystem,that is developing the distributed control subsystem into Agent. This is the microcosmic develop of Distributed Control System. The second direction is developing the control system into special MAS system. This is the macroscopical develop of Distributed Control System. Because of the different request for the designed goal and its circumstance,there are different needs for Agents in MAS system. So we divide Agents into different class according to their capability and the roles they acting on,which make up of two different kind of grading MAS system. Besides,we also have some research for the performance of Agent in each class and of the grading MAS system. Following the framework and managing model,we put forward the conception of grading MAS system and describe it as an data structure named "forest". We also put forward KCM model of Agent (the model composed of knowledge,capability and promise) and vindicate its affectivity by many experiments. Moreover,we introduce the ants system in bionics into grading MAS system ,try to study the abstract model of individual ant,the collective and the task and plot out the relief map and the process of solving the problem. We use the technology on Mobile-Agent to replace traditional technology on communication to meet the demand for the communication and evolution among Agents in grading MAS system. We also put forward three key problems should be solved in Mobile-Agent system and put up the achievement obtained in research.
|
PDF Full Text Request