Short-term power load forecasting based on DBO-TCN

Abstract Short-term power load forecasting is conducive to the stable operation of the power grid and can rationally allocate resources to reduce the waste of electric energy and promote the realization of the goal of “carbon peak and carbon neutral.” The current power load forecasting models generally have the disadvantages of poor accuracy and long prediction time. In this paper, we propose a dung beetle optimization algorithm combined with a time convolution network (DBO-TCN) model. First, we use DBO to optimize the parameters and then use TCN to predict the results and calculate the evaluation indexes as RMSE (Root Mean Squared Error)=3.7936; MSE(Mean Squared Error)=14.3916; MAE(Mean Absolute Error)=2.7448; MAPE(Mean Absolute Percentage Error)=0.016258; R^2(R-squared)=0.9833. The results are as follows: The model has high accuracy and good stability, which can improve the accuracy of short-term power load forecasting and maintain the stability.