ANN Optimization for Short Term Forecasting of Solar PV Power | Original Article

ABSTRACT:

One of the renewable energy resources used worldwide is solar energy. Its availability varies from day to day and even within a day itself depending on season. Due to this variability, forecasting of solar PV power becomes necessary but also challenging. The reasonably accurate forecast is an important input to the state load dispatch centres for scheduling power generation by various sources of energy. Artificial Neural Network (ANN) is widely used for such forecasting which involves training of the ANN using Bayesian Regularization and Levenberg-Marquardt algorithms. The forecasting approaches endeavour to minimize the error.  In this paper, ANN based short term PV power forecasting algorithm is used to forecast solar PV power 6 hour ahead  using trial and error approach and to compare different training approaches to reach at an optimized ANN solution.