Forecasting of power output of 2-Axis solar tracked PV systems using ensemble neural network

Photovoltaic (PV) based power generation system has been considered massively as one of renewable energy resource. However, the performance of PV system is sensitively affected by many factors including the weather and solar irradiance. The hybrid system is taken for solving this system output uncertainty. For improving the power management performance such this hybrid systems, the forecasting of power output of PV system has been proposed in some previous research. The precision of this forecasting has to be considered for building a high performed power management system especially for remote area where the very small power output is very important. Therefore, this paper proposes a novel approach of forecasting of power output of PV systems using ensemble neural network with four base forecasters. The PV system used in this research is equipped with 2-axis automated tracking with maximum output 10Wp. As base forecasters of ensemble structure, this research employs the multi-layer perceptron network with two hidden layer. According to the research results, the proposed method provides high accuracy prediction. Moreover, this method outperforms the individual MLPN based forecaster commonly used in the forecasting research.