Abstract: In order to improve the accuracy and efficiency of short-term power load forecasting, a load forecasting model named NRBO-TCN-GRU-Attention is proposed by combining the temporal convolutional network (TCN), gated recurrent unit (GRU) model, Newton-Raphson-based optimizer (NRBO) and the attention mechanism (attention). In this model, the NRBO is utilized to optimize the hyper-parameters. The TCN module extracts features from the load data and inputs the extracted features into the GRU module to capture the long-term dependencies in the load sequence. Next, the attention mechanism is utilized to reinforce the important features. Finally, the prediction results are output through the fully connected layer. The experimental results show that the proposed model outperforms other comparative models in four metrics on the two-day and one-week test sets, namely, coefficient of determination, meanabsolute error, meanabsolute percentage error and root-mean-square error, which validates the superiority and applicability of the proposed model.

Key words: short-term power load forecasting, temporal convolutional network (TCN), gated recurrent unit (GRU), Newton-Raphson-based optimizer (NRBO), attention mechanism