The Research On Load Forecasting And Energy Optimization Scheduling Of Micro-grid System

Distributed power generation using renewable energy has the characteristics of randomness and intermittence,which has a negative impact on the safe and stable operation of large power grids to a certain extent.The energy management of micro-grid can maintain the balance of system power,which was improved the reliability and economy of power supply.It is greatly significance that effective energy management of micro-grid to improve the reliable power supply and economic operation of micro-grid.In this thesis,the development of micro-grid has been discussed,the research status of micro-grid at home and abroad has been summarized,and the research progress in load forecasting and energy optimal dispatching of micro-grid has been analyzed in detail.Combined with the actual background,this thesis has expounded the significance of the research.Secondly,aimed at the influence of the volatility of the electric load on the energy management in the micro-grid,the electric load forecasting method has been proposed based on RBF neural network.In the selection of input data of neural network,the grey correlation analysis was used to select the training data.It was constructed that the preprocessing method of network input data,the determination of network structure and the determination of network output.The load forecasting model was constructed of micro-grid based on RBF neural network.Then,two improved combined forecasting models were proposed,which aimed at the problem that RBF neural network was easy to fall into local minimum in the process of forecasting.The first one was designed to reduce the non-stationarity of historical data.The EEMD is used to decompose the load time series,and the complexity of each sub-sequence was calculated by SE,which improved the forecasting efficiency.The second improved model was aimed at optimizing the structural parameters of RBF neural network.The data was optimized by GM(1,N)model to reduce the randomness of the parameters.The universal gravity search algorithm(GSA)was used to avoid the network falling into the local minimum problem.The memory function and group communication function of particle swarm optimization(PSO)were introduced into the velocity updating formula of the universal gravity optimization algorithm to avoid the phenomenon of continuous oscillation of the optimal value in the optimization process.Finally,based on the typical micro-grid system with photovoltaic,wind power,energy storage and power load,and considered the maintenance cost of energy storage unit,the maintenance cost of distributed generation unit,the cost of purchasing and selling electricity from large power grid,two kinds of energy optimal dispatching strategies were proposed of micro-grid,and the energy optimal dispatching model of micro-grid system in grid-connectedmode was constructed.The PSO,GSA and PSOSGSA algorithms were used to simulate the energy optimal dispatching models of different optimal dispatching strategies.The simulation results showed that the optimization results of PSOGSA algorithm were obviously better than those of PSO and GSA algorithm.It was verified that the economy of the optimal dispatching strategy by analyzing the operation cost of micro-grid under different dispatching strategies.