Optimizing one-hidden layer neural network design using Evolutionary Programming

This paper presents the optimization of one-hidden layer artificial neural network (ANN) design using evolutionary programming (EP) for predicting the energy output of a grid-connected photovoltaic system installed at Malaysian Energy Centre (PTM), Bangi, Malaysia. In this study, the architecture and training parameters of the multi-layer feedforward back-propagation ANN model had been optimized while the prediction performance of the ANN was maximized. The proposed evolutionary programming-ANN (EPANN) model employs solar radiation and ambient temperature as its inputs while the sole output is the kilowatt-hour energy output produced from the grid connected PV system. The prediction performance was quantified using the average correlation coefficient and it was maximized by determining the optimum values for the number of nodes in the hidden layer, momentum rate and learning rate during an evolutionary training. Besides searching for the optimal number of nodes and optimal training parameters for each model, the highest correlation coefficient for the prediction required for the EPANN was investigated. It was found that the maximum average correlation coefficient obtained for the EPANN training is 0.9962. Besides that, the testing process produced sufficiently high correlation coefficient value of 0.9976.