Adaptive Modularized Recurrent Neural Networks for Electric Load Forecasting

In order to provide more efficient and reliable power services than the traditional grid, it is necessary for smart grid accurately predict electric load. Recently, recurrent neural networks (RNNs) have attracted increasing attention in this task because can discover temporal correlation between current load data those long-ago through self-connection of hidden layer. Unfortunately, RNN prone vanishing or exploding gradient problem with increase memory depth, which leads degradation predictive accuracy. Many architectures address at expense complex internal structures increased network parameters. Motivated by this, article proposes two adaptive modularized RNNs tackle challenge, not only solve effectively a simple architecture, but also achieve better performance fewer parameters other popular RNNs.