A Digital Load Forecasting Method Based on Digital Twin and Improved GRU

Abstract

In view of the problems that most load forecasting methods are difficult to achieve high-precision data processing in the process of power grid digitization, a digital power grid load forecasting method based on digital twin and improved GRU is proposed. Firstly, the digital power grid model is constructed based on the digital twin technology, and the operation mode of the digital twin power grid is introduced in detail. Then, the time convolution network (TCN) and the gated recurrent unit (GRU) network are fused to design the GRU-TCN prediction model. The high-dimensional data features extracted by TCN are input into the GRU network for learning to enhance the model prediction performance. Finally, the GRU-TCN model is applied to the load forecasting of digital twin power grid, and the high-precision forecasting results are obtained. The experimental results based on the simulation platform show that the proposed method prediction results root mean square error is 28.428KW, and mean absolute percentage error values is 2.161%, which prediction effect is good.