A Rolling Prediction Method for Distribution Network Load Based on PID Control Theory

Load forecasting in distribution networks plays a crucial role in their operation. Addressing the limitations of data-based forecasting methods, such as data dependency and lack of interpretability, this paper proposes a tracking prediction method based on control theory. The stability and steady-state error of the proposed method are theoretically validated. Finally, the proposed method is compared with AutoRegressive Integrated Moving Average(ARIMA), AutoRegressive Distributed Lag(ARDL), and Long Short-Term Memory(LSTM) in terms of forecasting performance using actual load data. The results show that the proposed method achieves smaller forecasting errors and significantly reduces computation time compared to deep learning-based methods such as LSTM. This new approach provides a novel perspective for power load forecasting.