Dynamic risk assessment approach for deep foundation pit construction based on the integration of multisource data using D-S evidence theory

Construction of deep foundation pits is exposed to high risk considering the complexity of geological conditions and surrounding environments, and prompt risk assessment plays an important role in safe construction and management. A novel dynamic risk assessment approach based on the Dempster-Shafer (D-S) evidence theory is proposed to promote efficient assessment of construction risk with the multisource data of different stages. The risk index system is designed extensible and updated with the changing data types at different construction stages. With the D-S evidence theory and the measurement of Euclidean distance, the key factors including basic probability assignments and indicator weights are calculated dynamically. Based on the evaluated risk results, a novel risk control simulation algorithm is proposed to provide risk control suggestions for decision-making. The proposed approach was incorporated into the collapse risk assessment module of the construction information management system for Wuhan Metro Line 12. The risk assessments of Zhongyilu Station show the risks in the four construction stages were all Level I (negligible risk) and verified by the in situ conditions. The robustness of the presented approach was demonstrated by a following Monte Carlo simulation, and a case study of the collapse accident in Xianghu Station validated the feasibility of providing useful risk control suggestions. The presented approach facilitates dynamic and efficient risk evaluation of deep foundation excavation.