Author(s)

Lei Jin

Corresponding Author

Lei Jin

Abstract

In view of the increasingly prominent structural deformation risk of key nodes in urban subway under high-density operation and complex geological environment, this paper proposes and constructs a set of real-time monitoring and dynamic risk assessment technology system which integrates multi-source sensing, real-time transmission and intelligent analysis. By arranging fiber grating, laser ranging, inclination angle and microseismic sensors, a multi-dimensional monitoring network covering strain, displacement, vibration and other dimensions is formed, so as to realize minute-level real-time sensing of structural deformation of key nodes such as subway crossing tunnels and adjacent sections of deep foundation pits. Digital twinning and machine learning technology are introduced, structural mechanics model and convolutional neural network (CNN) algorithm are combined to extract key deformation features, and ARIMA/LSTM model is combined to realize short-term risk early warning. Furthermore, a dynamic risk assessment model coupled with real-time monitoring data and physical mechanism is proposed. Based on entropy weight -TOPSIS and Bayesian network, risk factors are dynamically weighted and failure probability is updated. A disaster chain propagation model is constructed to quantify risk consequences, and a reinforcement learning decision support module is integrated to realize risk response closed-loop management. The engineering case shows that the system can identify the sudden change of tunnel settlement 12 hours in advance and trigger high-risk warning, and successfully avoid major operational accidents, which verifies its engineering applicability and popularization value in the intelligent management and control of subway structure safety.

Keywords

Real-time Monitoring, Risk Assessment, Structural Deformation, Key Nodes, Urban Subway