Research On Multi-Agent Based Modeling Methods For Energy Complex Systems

With the high increase of global economy, the situation of energy safety and the conflict between energy supply and energy demand are increasingly severe. As a class of huge system with high complexity and openness, energy system is now becoming the focus of research. In the dissertation, the research of the energy complex system modeling problem starts with complex system modeling method. Therefore, the method of using Multi-Agent to model energy complex system is put forward based on the research of energy complex system because of the great advantages of Multi-Agent methods in dealing with the nonlinearity, dynamics, emergence and mutability of complex system.At first, the dynamic situation of research on all the methods of modeling energy system including our country's and foreign ones is analyzed in the dissertation, the comparison between Multi-Agent complex system modeling method and energy complex system is made to study their suitability. The features and common modeling methods of complex systems are further analyzed and studied based on energy system as a class of complex system. Secondly, the structure, type of various agents and the architecture, interaction and communication of a Multi-Agent system are studied. On the above basis, energy complex system is analyzed to build the hierarchy model of energy complex system based on Multi-Agent. Further, various agents including energy supply agent, energy demand agent, energy prediction agent, energy coordination agent, energy decision agent and energy region agent are built according to the targets and tasks that energy complex system should accomplish and all their structures and methods of implementing are designed. Meanwhile, on the VC++development platform the interactive interfaces of energy complex system based on Multi-Agent and every agent and the functions they should realize are designed and developed. Based on the research of collaboration and communication mechanism of Multi-Agent, the specific communication structure based on XML and KQML language is proposed, which is suitable to the energy complex system.At the end, using theε-SVR method an energy prediction agent is built and simulated by Matlab to realize prediction, which shows the validity and social value of the research on energy complex system modeling based on Multi-Agent and sets a great theoretical basis for its further implementation.