Abstract: The process of diagnosing tunnel health as a fuzzy system has the characteristics of randomness and fuzziness. Traditional risk assessment methodologies do not account for the fuzziness and randomness of risk. Therefore, using cloud theory, a new comprehensive risk assessment model was proposed for shield tunnels aimed at the fuzziness and randomness of health diagnosis system of shield tunnel structure. Firstly, the normal cloud model of evaluation terms is constructed, which reflects the health status of the tunnel structure. Then, the normalized value of monitoring data of tunnel structural health status index is transformed into membership cloud model by the principle of the reverse cloud generator. And the importance linguistic value of the health status index is transformed into the weight of the cloud which is used for characterization importance of health index. Finally, the cloud theory computing method is used to diagnose the health status of the tunnel structure, and the cloud model of health status is obtained. The improved method of cloud model was applied to diagnose the structural health status of shield tunnel of Metro Line 2, and this method greatly improves the visualization and robustness of the diagnosis results.