Aggregated Power Prediction of Thermostatically Controlled Loads Based on Multi-scale Residual Gated Recurrent Unit Network

Residential Thermostatically Controlled Loads (TCLs) takes a large proportion in the power grid, but the distributive characteristics would make it difficult for dispatch centers to directly get their aggregated power. In order to get TCLs to participate in the power grid operation, this paper proposed a multi-scale residual gate recurrent unit network to predict the aggregated power of TCLs. This method first extracts multi-scale feature information by multi-scale convolution, then uses residual learning to deepen the network structure to further improve the performance of the algorithm, and finally uses the gating mechanism of gated loop unit network to update and maintain higher-order semantic feature information. The experimental results show that the prediction model proposed in this paper has better prediction effect.