Forecasting Models of Electricity Demand and Renewable Energy Generation for Virtual Power Plant

The renewable energy utilization has been an indispensable choice to ensure secure and sustainable energy supply. However, the high penetration of renewable energy integration will lead to increase the power grid disturbances. In this research, short-term forecasting method for renewable energy power generation and electric demand is proposed to achieve better energy management. A forward neural network forecasting technique is used with 4 initial variables (irradiation, irradiation changes at different times, load demand, and wind speed) and yields 15 forecast models for each season. Forecasting models are taught and tested by MATLAB programming. The test results shows that the lowest mean absolute percentage errors (MAPE) of PV power generation, wind power generation and the electric demand are 12.41 %, 16.17 %, 6.29 % respectively.