Abstract: On July 20, 2021, the heavy rain in Henan Province attracted wide attention due to its heavy rainfall intensity, long duration and frequent floods. The heavy rain caused serious waterlogging in Zhengzhou, Kaifeng and other cities. Aiming at the problem of urban waterlogging risk assessment, the data of 12 automatic monitoring stations were screened, and a semi-structural index system consisting of two dimensions of short-term risk and long-term risk and nine influencing factors was constructed. In the input layer, information entropy weight vector was adopted, and in the criterion layer, three weighting methods were adopted: entropy weight vector, equal weight vector and decision preference weight vector, which constitute the improved risk fuzzy evaluation model. Waterlogging risk was assessed at five levels (very high, high, medium, low, and very low) for 12 sites. The results show that: in the short-term risk assessment, site 5 means very high risk, and site 10 is medium risk. In the long-term risk assessment, site 6 is high risk, and site 8 is very low risk. In the final evaluation of the target layer, the three methods all show that the risk of No. 5 and No. 2 sites is very high, and the risk of No. 7 and No. 8 sites is low. Compared with the actual situation of waterlogging in Kaifeng City, the evaluation system and improved model are considered to have strong applicability. Accurate evaluation results of short-term risk, long-term risk and comprehensive risk can be provided for urban waterlogging risk, providing reliable data support for urban drainage pipe network improvement and waterlogging prevention and control. It provides a basis for making emergency disaster reduction plan and countermeasures to deal with waterlogging caused by rainstorm.