Multi-source data fusion early warning method for 3e risks in building systems

Abstract

The invention discloses a multi-source data fusion early warning method for 3E risks in building systems, relating to the technical field of construction engineering. The invention comprises collecting and preprocessing heterogeneous data from engineering technical risk, energy, and environment; aligning multi-time-scale data and constructing a fusion matrix; extracting risk coupling features through heterogeneous graph convolution; dynamically quantifying comprehensive risk coupling strength; adaptively generating multi-level early warning thresholds; performing risk tracing and hierarchical response decisions; and online incremental updating of models and parameters. By designing a multi-time-scale feature alignment and fusion mechanism and a dynamic coupling risk assessment model, the invention enables lossless collaborative analysis of three types of asynchronous monitoring data of engineering technical risk, energy, and environment, and more accurately quantifies the comprehensive risk strength under their interactions.