Research on identification of control delay times in centralized heating systems using dynamic EWMA-sliding time window method

Centralized heating systems exhibit significant control delay characteristics, which hinder timely achievement of control objectives and prevent prompt adjustment of heat supply according to demand, resulting in unnecessary energy waste. To support the mitigation of time-delay impacts and facilitate enhancements in energy efficiency and user comfort, this study proposes a dynamic exponentially weighted moving average -sliding time window method that automatically identifies control delay times through historical operational data analysis. To validate the detection accuracy and noise robustness of this approach, comparative applications were conducted with the existing dynamic cross-correlation function method. Results demonstrate that compared with the dynamic cross-correlation function method, the proposed method achieves an 82.4 % reduction in mean absolute error (MAE) of detection accuracy and a 74.7 % improvement in noise robustness. The methodology was subsequently applied to historical operational data from the primary and secondary networks of FXJY heat exchange station in KF's centralized heating system, revealing control delay times of 2 h for the primary network and 7.2 min for the secondary network. These findings provide critical time-delay parameters for predictive control optimization in heating systems, while the proposed methodological framework and empirical conclusions offer theoretical support and practical guidance for smart heating management.