Based on the interval analytic hierarchy(IAHP)method and the extended approximation ideal solution sequencing(TOPSIS)method, a reasonable real-time security risk identification model was established. The proposed model was employed in the risk analysis of a large-scale water diversion project near the urban tunnel section. The multiple-index decisions were made on the number of safety risk intervals at different construction stages. According to the engineering data of the corresponding stage, the safety risk factors are sorted. The feasibility and effectiveness of the real-time security risk identification model was tested.