Solar Power Generation Forecasting by using Artificial Neural Network and Back Propagation Algorithm

Accurate forecasting leads to minimal losses and maximum profit for Power Sector Companies. Hence, Forecasting is important for Power Sectors Company for bidding, coordination, planning, control and operation of other sources. The peaks demands on power system are met by alternate source of energy mostly diesel generator and non-conventional source of energy. Out of there two power generation predictions from non-conventional source of energy is very important since it is nature dependent and non-liner. Forecasting based on non-linear data itself is a challenging task and developing a model is also challenging. In this work machine learning based solar power generation forecasting model is presented which can predict non-linear data based on its training algorithm. Artificial Neural Network (ANN) with single hidden layer and Back propagation algorithm (BP) with multiple hidden layer is presented for a week ahead Solar Power Generation Forecasting (SPGF), and the results shows the machine learning with more number of hidden layers will give better forecasting accuracy, i.e. BP algorithm exhibit better performance over single layer Neural Network.