Multi-agent System And Its Research In The Micro-grid

As systems becoming more and more complex, they require higher and higher intelligence,systems are gradually changing from being centralized controlled based on the traditional precise mathematic models to being distributed controlled based on artificial intelligence. Multi-agent technology as an effective way to solve complex distributed problems are being paid more and more attention, and researchers also conduct in-depth study about it.Multi-agent is a complex technology, involving several research fields, although it is successfully applied in some aspects, in itself is not perfect, there are many problems should to be further researched. In this paper, the collaborative learning of Multi-agent system is researched, and the reinforcement learning method is introduced to the Multi-agent collaborative learning. By researching and improving the Q-learning algorithm of reinforcement learning, to enhance the Multi-agent system's performance on learning and collaboration. Furthermore, this paper also studies about the Multi-agent system structural model, by introducing knowledge-sharing mechanism, facilitator concept and proposing team structure idea, it can minimize spatial data of Multi-agent system, reduce complexity of communication between Multi-agent, and optimize the organizational structure of multi-Agent system, ultimately, a new structural model of Multi-agent system based team ideal is proposed, and simulation shows the progressive nature of the new structural model.Micro-grid is also a hot research area, especially today's climate problems gradually become to an international hot issues, the world's major countries have put forward plans of developing clean energy projects based distributed generation, namely to develop micro-grid of smart grid projects, micro-grid has also been listed as key basic research by nations. But there are many key issues are not solved, such as how to control running it, how to control the micro power grid splitting, and network impact when parallel in a large grid and so on. In this paper, it takes micro-grid as a carrier in order to study the control performance of Multi-agent system. This paper applies the result of study on Multi-agent system to control micro-grid voltage and reactive power, and propose three layers structure of Multi-agent system to control it, the role of each agent and detailed control strategy are given, lastly, the simulation shows the micro-grid voltage and reactive power can be effectively controlled by the Multi-agent system.