Hybrid wavenet model for short term electrical load forecasting

Precise electric load forecasting is very crucial in achieving better cost effective risk management plans. This can be achieved by a Short Term Electricity Load Forecasting (STLF) model. This paper proposed a Cascaded Feed-Forward BPN wavenet forecast model to perform the STLF. The model is composed of several neural networks along with wavelet transform. The historical electricity load data is processed using a wavelet transform technique. This load data is decomposed into several wavelet coefficients using the discrete wavelet transform (DWT). The several wavelet coefficients are then used to train the neural networks (NNs) and later, used as the inputs to the NNs for precise electricity load prediction. The Levenberg-Marquardt (LM) algorithm is selected as the training algorithm for the NNs. To obtain the final forecast, the outputs from the NNs are recombined using the same wavelet transform technique.