Development of an Elman Recurrent Neural Network Model for Short-Term Maharashtra State Electrical Power Load Prediction with Special Emphasis on Seasonal Changes

In this paper, the parameter-wise optimization training process is implemented to achieve an optimal configuration of Elman and Jordan recurrent neural network (RNN) models. The aim is to examine the prediction ability of the proposed models to predict one--day-ahead electric power load simultaneously as usual to appose 1–24 hour forecast in sequel with a special emphasis on seasonal changes over the year. Various learning algorithms are used to accelerate the training of neural networks (NNs). Experimental results indicate that the Elman neural network clearly outperformed the Jordan network in various performance metrics such as mean square error (MSE), normalized MSE, correlation coefficient(r) and mean absolute percentage error (MAPE) during evaluation process. Empirical results show that the proposed RNN model is consistently performs well on daily and on monthly average basis in terms of prediction accuracy. The proposed method gives acceptable errors in all seasons, months and on daily basis. The average prediction error in the year 2006 is as low as 2.06%.