Prediction of Output Solar Power Generation using Neural Network Time Series Method

Abstract

Artificial neural networks (ANN) are used for many years to optimize the results of various problems in various sectors and disciplines like, Engineering, Industrial applications, Finance, Medical applications, Economy, Forecasts, etc. The training ability of ANN has capability to deal with nonlinear and complicated issues termed for its utilization to solve projection troubles. In this paper we have developed a auto regressive nonlinear or NARX structure with external input, for the forecasting of AC system output of a solar power plant placed in RCEW college jaipur (Rajasthan) for the year 2014. In this system for the purpose of forecasting, the Levenberg-Marquardt (LM) optimization approach was utilized as it aids the finest training rate pursued as a back propagation method for the multilayer feed forward artificial neural networks architecture using MATLAB® R2013 ANN time series toolbox. The outcome of the model concluded that the preferred procedure is powerful in prediction of forthcoming future energy generation demands for the daily operational planning of solar power generation.