Application of Building Information Modeling in Subway Foundation Pit Support Risk Management

Abstract

To address issues such as low safety factors and difficulty in predicting the force variation of support systems in subway station foundation pit construction, this study develops a subway foundation pit support risk management system based on a risk data monitoring method combining Building Information Modeling technology and gated recurrent neural networks. In constructing the risk management system, a fuzzy analytic hierarchy process model is also used to assess and analyze the foundation pit risks. Experimental results show that the hybrid risk data monitoring method records a displacement of 0.28mm at monitoring point 6, while other methods—Cooperative Game Empowerment, Grey Relational Analysis, and Relativistic Conditional Generative Adversarial Network —record displacements of 0.16mm, -0.21mm, and 0.17mm, respectively. Additionally, empirical analysis of the constructed foundation pit support risk management system reveals that the hybrid subway foundation pit support risk management system measures a settlement value of -10.83mm, while the Generative Adversarial Network and Grey Relational Analysis-based systems measure maximum settlements of -10.11mm and -9.23mm, respectively, all of which are less accurate than the hybrid risk management system. These results demonstrate that the hybrid subway foundation pit support risk management system performs well in risk control and monitoring, ensuring the stable operation of subway construction. This research contributes to future control of foundation pit excavation support safety and the improvement of dynamic risk management levels.