Development of a Neural Network Based PV Power Output Prediction Application Using Reduced Features and Tansig Activation Function

Abstract

Many research works were done on the prediction of this PV power using Artificial Neural Network and other Artificial Intelligence method to somehow address some issues encountered in injecting the excess power of a Grid Connected Solar PV system into the grid. However, it was observed that simulation processes posed a cumbersome task. Thus, this paper attempts to develop PV power output prediction application software. It presents a methodology used in developing a NN - based PV-output prediction application with reduced features. It integrates techniques used in previous studies and utilizes some common Software applications such as Soft Computing tools, spreadsheets and IDE in the application development. As a result, the developed application software was able to perform PV output forecasting easily and dynamically.