Abstract: Currently, construction risk prevention and regulation are difficult in expansive newtype underground spaces. As a result, the term of large urban underground space (ULUS) is defined, and the characteristics of the ULUS construction risk are examined. Then, the idea of investigating the construction risk of the ULUS from a systematic perspective is advanced, splitting the risk system into three parts: structural, environmental, and organization and management. Based on the Analysis of each subsystem′s characteristics and the relationship between the subsystems, a multifactor coupling and evolution model of major risks in the ULUS construction is established by combining the Bayesian network, fuzzy comprehensive evaluation method, Markov chain, and system dynamics. Furthermore, the concept of the coupling coefficient of risk factors is proposed in the model of a Bayesian network framework, which can quantitatively express the coupling effect of risk factors while streamlining the creation of the Bayesian network.  Furthermore, the Markov chain and system dynamics forecast the growth behavior of the environment and the organization and management subsystems to predict the risk evolution trend during the construction process. Finally, the model also evaluates the risk associated with the building of the Dongdaqiao station on Beijing metro line 17. The findings demonstrate that the construction risk of the ULUS has clear stage characteristics, and the risk reaches the greatest in the excavation stage; the coupling effect of the risk factors has a significant impact on the likelihood of risk events, ignoring the coupling effects of risk factors will cause higher deviations in the risk analysis results.

Key words: urban large underground space, major construction risks, multifactor coupling, risk evolution, risk factor coupling coefficient; Bayesian network