Digital twin-enabled anomaly detection for built asset monitoring in operation and maintenance

Effective asset management plays a significant role in delivering the functionality and serviceability of buildings. However, there is a lack of efficient strategies and comprehensive approaches for managing assets and their associated data that can help to monitor, detect, record, and communicate operation and maintenance (O&M) issues. With the importance of Digital Twin (DT) concepts being proven in the architecture, engineering, construction and facility management (AEC/FM) sectors, a DT-enabled anomaly detection system for asset monitoring and its data integration method based on extended industry foundation classes (IFC) in daily O&M management are provided in this study. This paper presents a novel IFC-based data structure, using which a set of monitoring data that carries diagnostic information on the operational condition of assets is extracted from building DTs. Considering that assets run under changing loads determined by human demands, a Bayesian change point detection methodology that handles the contextual features of operational data is adopted to identify and filter contextual anomalies through cross-referencing with external operation information. Using the centrifugal pumps in the heating, ventilation and air-cooling (HVAC) system as a case study, the results indicate and prove that the novel DT-based anomaly detection process flow realizes a continuous anomaly detection of pumps, which contributes to efficient and automated asset monitoring in O&M.