One Day Ahead Prediction of PV Power Plant for Energy Management System Using Neural Network

One of the aims of this paper is to reduce electricity bills and provide added value to its users, besides producing environmentally friendly and emissions-free electricity. Therefore, the power generated from solar panels must be maximized to achieve the target expected by the user. In this study, the author means to predict power output using Elman Neural Network (ENN) and Feed Forward Neural Network (FFNN). The Levenberg-Marquardt algorithm is used for the learning process as an activation function of the time series of PV power production, and the inputs used in Neural Networks (NN) are several meteorological variables. Some error values from the simulation results will be evaluated to estimate the accuracy of the forecasting method, for energy management systems and those with the smallest MSE error values will be selected as a reference in planning the establishment of solar panel power plants in the future. The results show that FFNN provides the best MSE, for all cases an example in case II days 3 and 4 are 0.01471% and 0.00097% respectively.