Energy Forecasting For Grid Connected MW Range Solar PV System

An accurate forecasting of PV output is essential to improve real-time control performance and to reduce possible negative impacts. For an energy management system (EMS) of distributed energy resources, accurate forecasting of solar irradiation and thus PV power output can reduce the impact of uncertainty for PV power generation, improve system reliability, and increase the penetration level of the PV power generation system. Solar power output can be predicted based on historical solar irradiance and weather data, using Artificial Neural Networks and then is converted to PV power output. Neural Networks is designed to train, visualize, and validate network models using feed forward network with back propagation algorithm. The model is trained; using data collected from a 10 kW PV system which includes beam solar irradiance, relative humidity, temperature and wind speed as input for the training set. The future DC and AC power outputs are predicted for any given day. Also the power forecasting is being extended for a 100 kW and 1 MW PV system.