Power Forecasting And Real-time Energy Management Research Of Photovoltaic Plant With Battery Storage

Because of the impact of geography, climate, weather and other external factors, the output energy of photovoltaic plant is intermittent and unpredictable, which will cause the complexity of energy exchange between the photovoltaic plant and the large power system. The massive photovoltaic plant accession to the large power system will affect the security and stability of the power system. Further, the photovoltaic plant can buy electricity from the power company and can also sold surplus energy of the own PV to power companies or provide emergency power support service. The basic analysis theory and design methods of energy management of 1MW photovoltaic plant in Qinghai are studied, including power generation forecasting, energy management of photovoltaic plant.With the increase of the capacity of photovoltaic generated systems, power forecasting of PV system plays a very important role in optimal combination of economic dispatch and optimal power flow. In this thesis, prediction models and design methods of the photovoltaic power generation based on Elman is proposed. The meteorological factors which influence power output of photovoltaic power generation system is analyzed and the measures to treat and identify these factors are given. And the design method of input layer, hidden layer and output layer of Elman neural network is given. Compared with BP neural network and NSET, the prediction accuracy of the Elman neural network model is better.The power output of PV will vary with external conditions so that they cannot satisfy the load independently and energy storage device are required to provide support and backup. Energy management system of photovoltaic plant becomes significant for the stability and economic operation as the power flow between PV units, energy storage units, the grid and the load need to be optimally controlled and managed. Real-time energy management for grid-connected photovoltaic plant is proposed. It previously divides the 24 hours in a day into peak time, normal time and valley time. Different dispatching strategies are used according to current time and the SOC of battery storage. Study results indicate the proposed method cannot only realize economical operation, but also help outer grid with “Peak clipping and valley filling”.