Hybrid Artificial Neural Network with Meta-heuristics for Grid-Connected Photovoltaic System Output Prediction

Abstract

This paper presents the performance evaluation of hybrid Artificial Neural Network (ANN) model with selected meta-heuristics for predicting the AC output power fof a Grid-Connected Photovoltaic (GCPV). The ANN has been hybridized with three meta-heuristics, i.e. Cuckoo Search Algorithm (CSA), Evolutionary Programming (EP) and Firefly Algorithm (FA) separately. These meta-heuristics were used to optimize the number of neurons, learning rate and momentum rate such that the Root Mean Square Error (RMSE) of the prediction was minimized during the ANN training process. The results showed that CSA had outperformed EP and FA in producing the lowest RMSE. Later, Mutated Cuckoo Search Algorithm (MCSA) was introduced by incorporating Gaussian mutation operator in the conventional CSA. Further investigation showed that MSCA performed better prediction when compared with the conventional CSA in terms of RMSE and computation time.