Methodology for building a digital twin of a capital construction facility based on real-time multisensory monitoring and construction control

Abstract

The digitalization of the construction industry is shifting from static BIM models to cyber-physical systems capable of synchronizing the virtual representation of a capital construction object with its actual state in real time, however, this is hindered by discreteness of control, heterogeneity of data sources and delays in the transmission of telemetry, critical at the construction stage, when the risk of irreversible deformations is maximum. The purpose of the study is to develop a methodology for building a digital twin that integrates multisensory monitoring and construction control for profit , with minimal latency and high data reliability. The construction site of a high—rise complex with a reinforced concrete frame was used as an experimental base; the digital basis is LOD 400 BIM with IFC exchange. Data collection was provided by a distributed network (124 string strain gauges, 45 high-precision inclinometers, 20 three-axis accelerometers; more than 500 measurement points with a frequency of 1-100 Hz), supplemented by ground-based laser scanning and UAV photogrammetry; over 14 months, over 15 TB of information has been accumulated. The processing included noise filtering, registration of point clouds in the geodetic coordinate system, automated detection of geometric deviations and cross-verification using independent measurements (including a total station), as well as algorithmic compensation for external influences (temperature drift, site interference). The results showed a significant heterogeneity in the quality of telemetry over communication channels (variability in latency and packet loss), revealed excess tolerances for vector deviations of individual columns, and recorded overspending of concrete in grippers (up to ~7-8%) with correlation with formwork deformations detected by scanning data; temperature corrections changed the interpretation of deformations by tens of percent, confirming the need for dynamic adjustments. The discussion proves that the value of a digital twin is determined not by the number of sensors, but by the quality of mathematical models for integrating and filtering heterogeneous flows, which reduce false alarms and switch control to predictive control mode. The practical conclusion is the expediency of a multi—level architecture sensors — primary processing - analytical core in BIM , suitable for scaling and further intellectualization of construction processes.