In view of ambiguity and randomness in the risk assessment of urban underground diseases, a risk assessment model of urban underground diseases based on cloud theory and set pair theory was proposed to quantitatively assess the disease risk. First, according to the relevant standards, the 25 assessment indicators of disease risk were selected from both the probability of risk occurrence and the risk consequences; a four-level risk assessment system and an index quantification standard were established. Second, by introducing the cloud theory in the field of uncertainty artificial intelligence studying the relationship between ambiguity and randomness, assessment indicators were transformed into cloud indicator according to the inverse Gauss cloud algorithm involving the second-order and the fourth-order center distance. Third, comprehensively considering the subjective cloud weight and the objective critic weight, a coupling cloud-critic weight optimized by the least squares method was proposed to distribute weights for cloud indicators, andthe least square method is utilized to combine and optimize the two kinds of weights. After a step-by-step calculation, the cloud indicator can be transformed into a comprehensive cloud. Final, a similarity calculation method of cloud model was proposed based on the set pair potential and the "3En" rule of Gauss cloud model, by introducing the set pair analysis (SPA) theory. The "3En" rule of Gauss cloud model counting cloud droplets and performing similarities, differences and counter analysis, and the set pair potential calculating membership degree of comprehensive cloud were used to judge the risk level of disease. Quantitative evaluation of Guiyang road detection engineering project has been carried out utilizing the established model. The evaluation results agreed well with the actual conditions of excavation verification and verified the applicability of the model. In addition, the model was compared with the evaluation model constructed by the analytic hierarchy process, fuzzy comprehensive evaluation and fuzzy analytic hierarchy process method; the results showed that the proposed model in this study achieved the higher evaluation accuracy when compared with the other models. It is thus feasible to apply the proposed model to the risk assessment of urban underground diseases.
