Research On Energy Management Strategys In Micro Grid Based On Multi-Agent-System

As the global energy crisis and environmental problems are becoming more serious, much attention has been paid to renewable energy generation such as wind power, solar power, etc. Safety stability problems will be likely to occur if the distributed generation(DG) is directly connected to the power grid.. In order to make full use of renewable energy generation, the microgrid(MG) is generated in the field of distributed generation. Microgrid is composed of a variety of distributed generations, energy storage system and loads. Microgrid is an independent and controllable system and can achieve flexible conversion between the grid-connected mode and the island mode. If the microgrid can be managed effectively, the reliability, stability and economy of power supply will be improved effectively. The microgrid energy management and strategy is one of the core problems in microgrid research. This thesis has carried out some research works about microgrid energy management and strategies as follows:(1) Microgrid energy management based on forecasting of load and renewable energy generation has been researched. First of all, the bi-level model of microgrid has been built based on multi-agent system. Then, the objective function which minimizes the operating cost, fuel cost and environment loss cost, is given. Using a dynamic segmentation penalty function method will convert the constrained optimization problem to the unconstrained optimization problem. Using binary contrast to determine the weight coefficient of the objective function, the selection of weight coefficient is more scientific and rigorous. In addition, the fuzzy neural network(FNN) prediction has been used for forecasting and optimized by particle swarm optimization(PSO) algorithm. According to the forecasting data, the objective function can be optimized by central force optimization(CFO) algorithm and improved particle swarm optimization(IPSO) algorithm, solving the micro grid optimal dispatch combination. The simulation results of the output powers of distributed generations can be got by two kinds of optimization algorithms. The two kinds of simulation results can be compared, and the comparing result demonstrates the validity of proposed algorithm.(2) Studies on bidding strategies of distributed generators in microgrid electricity market based on game theory have been conducted. First, a bi-level multi agent system(MAS) is structured. Then, a bi-level day-ahead bidding model is formulated for the bidding optimization of distributed generation and the bi-level bidding model is transformed into a mixed integer linear program(MILP) from a mathematical program with equilibrium constraints(MPEC) with binary expansion of quantity-price bids. Then formulation of an equilibrium problem with equilibrium constraints(EPEC) is processed for all distributed generation units to obtain the Nash equilibrium solution based on game theory. An auxiliary optimization problem is also formulated to find the unique Nash equilibrium for bidding game in consideration of maximizing the transaction price. Both non-cooperative and cooperative bidding games are addressed by means of the proposed energy management approach in the designed bi-level MAS market finally, and the simulation results demonstrate the feasibility of proposed bidding strategies.(3) The last chapter focuses on microgrid energy management in grid-connected model based on multi-time scale(MTS) theory. Firstly, a three-level multi-agent system is constructed and interactions among agents. Secondly, objective functions can be established and bidding strategies in power markets are formulated. Microgrid’s bidding powers and electricity prices can be got by nash equilibrium. Thirdly, under the two kinds of time scales, the daily schedule and real-time schedule are built for distributed generation power output optimization. The objective function and constraint conditions are built respectively in daily schedule and real-time schedule. The daily schedule can provide distributed generation output and start-stop schedule to real time schedule, and according to the real-time electricity price to adjust load. The two time scales perform theirs own functions and support each other. Afterwords, chaos theory based on Lyapunov index prediction model is used in the short-term forecast for load and wind turbine. According to the forecast data, the optimal allocation of the output power of distributed generation is achieved by central force optimization algorithm. Finally, the simulation results demonstrate that the effectiveness of the system energy management and strategies.