Identification and evaluation of operation regulation strategies in district heating substations based on an unsupervised data mining method

Abstract

In district heating (DH) substations, the operation regulation strategy has a significant impact on heating-energy consumption. To effectively diagnose and optimize the operation of DH substations, existing regulation strategies should be quantitatively identified and evaluated utilizing historical heating operation data. However, the individualization of regulation strategy caused by different building heat demand poses a challenge to the reverse identification. This paper first introduces the parametrically represented regulation strategies of DH substations, followed by a reverse identification of these strategies using an unsupervised data mining method based on Gaussian mixture model (GMM) clustering. Finally, an index of equivalent supply–demand matching coefficients (ESDMCs) is proposed to evaluate the operation effect and diagnose inefficiency in regulation strategies. To demonstrate its applicability, the proposed method was used to analyse the historical operation data obtained from a DH system in Tianjin, China. The results reveal that GMM clustering can be used to effectively identify individual regulation strategies during a heating season, thereby providing insights into the operational optimization of DH substations using the ESDMC evaluation index.