Research On Intelligent Control System Based On Multi-agent System

With the development of technology, the task of control is becoming more and more complex and multifold, and traditional intelligent control theories show some disadvantages in both design of control system and effect of control. Multi-agent technology that develops from DAI (distributed artificial intelligence) offers a new approach to intelligent control in complex system with large scale.Instead of going in for a single,huge system, the Multi-agent control system achieves final control of system by cooperating of several agents with different functions. Multi-agent control system overcomes the difficulty of knowledge management when the knowledge base is huge, and makes the system more robust, reliable and self-organizing. On the other hand, it also reduces the complex degree of system design.In this thesis, the integration of multi-agent system and intelligent control theory, what is called MAS control system, is first studied. Then, the mechanism of communication and cooperation of MAS control system is analyzed. Finally, a structure of MAS control system is presented based on blackboard structure. By an example, the idea of multi-agent theory and application in intelligent control is further illustrated, and the experimental results indicate that it finally realizes the control of system by cooperation of agents.Finally, this thesis researches the learning mechanism and knowledge acquisition of multi-agent system. It applies rough set, which is a new soft computation method, in knowledge acquisition of intelligent control system, and a method of the knowledge acquisition of intelligent control system in the large scale knowledgebase is proposed, which can overcome the defect of slowly extracting long rules.