Deep Neural Network based Forecasting of Short-Term Solar Photovoltaic Power output

Renewable energy integration to the conventional power grid is a challenge and requires an accurate forecasting of power output from the renewable energy sources for ensuring the reliability and grid stability. Many forecasting techniques for different time horizons were developed using different machine learning techniques. In the recent past mostly forecasting techniques based on artificial neural networks were developed. But, looking at the environmental parameters like insolation, temperature, sky clearness index and cloud cover etc., and its variable behavior makes the forecasting more complex. To address., complex and non-linearity issues in many applications, deep neural networks were proved effective and hence an attempt made in this paper forecasting power from solar photovoltaic plant for very short-term durations through deep neural networks model and compared the same with ANN model with only one hidden layer and found significant improved accuracy in deep neural networks.