Short-term electricity load forecasting method based on multilayered self-normalizing GRU network

A multilayered self-normalizing gated recurrent units (MS-GRU) model is proposed for short-term electricity load forecasting. This model introduces scaled exponential linear units (SELU) activation function to squash the hidden states to calculate the output of the model. Meanwhile, the squashed states also contribute to the calculation of update gate, reset gate and candidate state of GRU. Therefore, exploding and vanishing gradient problem can be overcome in a stacked GRU neural network using this self-normalizing method. Fuzzy cluster-means (FCM) algorithm is used for the selection of similar days of electricity loads. Experiments illustrates that the MS-GRU model can give a more accurate forecast to the short-term electricity load compared with other RNN models.