Towards automated fault detection and diagnosis in district heating customers: generation and analysis of a labeled dataset with ground truth

This study aims to develop a framework for automated fault detection and diagnosis (AFDD) in district heating (DH) substations by comprehensively understanding typical faults. AFDD is presently dependent on manual detection and diagnosis, leading to limitations. To address this issue, the study utilized data from 158 fault reports and smart heat meter data from residential buildings in Denmark to investigate common faults and conduct a fault impact assessment. The study suggests additional indicators for use by DH utility companies to detect anomalies in the future. The findings indicate that greater attention to fault detection and diagnosis can decrease energy usage and return temperatures, demonstrating the significance of AFDD implementation.