An adaptive wavelet neural network-based energy consumption forecasting of metallurgical corporation

A novel approach -adaptive wavelet neural network (AWNN) is proposed in this paper for energy consumption forecasting. The commonly used Morlet wavelet has been chosen as the activation function for hidden-layer neurons of feed-forward neural network (FFNN) which has been trained by hierarchical genetic algorithm combined with gradient descent algorithm. To demonstrate the effectiveness of the proposed approach, month-ahead prediction of energy consumption in Zhu Jiang Iron and Steel Groupcorporation is considered. The forecasted results clearly show that AWNN has better prediction properties and needs less learning time compared with WNN.