| As a new generation and distribution electric power system, Microgird adopts distributed generation as its major power source; therefore its secure and stable operation and high-quality power supply highly depend on a perfect Microgrid Energy Management System, namely MEMS. Distributed generation is characterized by dispersibility, diversity, environmental uncertainty, controlling complexity and so forth, which generates huge challenges to energy management of microgrid. At the demanding requirements of energy availability, power quality and economical efficiency, the conventional Centralized Controlling Energy Management System can no longer accord with the demands. From a perspective of technology application of multi-agent distributed artificial intelligence, this thesis constructs a Multi-agent Microgird Energy Management System on basis of JADE development form, and its specifics are as follows:Firstly, This paper researched the microgrid which is oriented to the renewable energy and deeply analysed the characteristics of the microgrid.The architecture and function of MEMS are designed.Combined with the openness,dynamic,hierarchical and nonliner characterized micro-grid energy management and the theory of Multi-agent technology,the feasibility of applying Multi-agent technology to MEMS is also analysed.Furthermore, the critical issues are discussed for the Multi-agent energy management.Secondly, the mutli-agent MEMS model is established based on the Agent-Based Modeling and Simulation(ABMS) method by top-down analysis and micro-macro modeling. To begin with, the basic Element Agents, such as Wind Turbine Agent, Photovoltaic Agent, Battery Agent, Diesel Engine Agent and Load Agent, are mapped as the power supply physical nodes for the micro agent modeling. Furthermore, in order to increase the system openness and extensibility, the generality of the physical node is refined and the Function Agent is designed in accordance with the requirements of microgrid energy management, such as Optimization Agent based on genetic algorithm, Alert Agent for the power quality evaluation based on fuzzy theory, Data Process Agent for the interface management etc. Then throughout the characteristics of MEMS and the demand for optimization and coordination control, a Multi Agent System, namely MAS, is proposed by drawing from the management mode in the work and life of mankind. The MAS model is consisted of three layers, that is individual layer, organizational layer and social layer, and it has good autonomy, openness, dynamic and efficiency.Thirdly, the microgird management strategy is researched, energy management strategy and its realization of MAS cooperation mechanism for both the grid-tied and island microgrid are put forward. The grid-tied operation of microgrid energy management strategy is discussed in the normal operation and peak shaving request of the grid respectively. Moreover, when the microgrid operates in offline mode, particularly for the island mode, the energy management strategy is a kind of multiple time scale plan composed of the day-ahead schedule, ultra-short-term schedule and real-time schedule. Against the requirements and characteristics of various time scales, this paper puts forward the objective function and constraint conditions for the corresponding time scale and adopts the genetic algorithm to realize the optimization of various time scales.Finally, synthesizing the research contents above, the paper preliminary develops a distributed MEMS oriented to the renewable energy on the JADE platform with the eclipse software and Mysql database. The distributed real-time data acquisition, on-site processing and display are also achieved by MEMS. Moreover, the power quality assessment and monitoring, the optimization and scheduling are realized. Finally, a micro grid paradigm is adopted for the experimental verification, and the results show that multi agent applied to the micogrid energy management system is verified feasible and effective. |
PDF Full Text Request