Risk Assessment and Dominant-Factor Identification for Pipe-Joint Misalignment in Pipe Jacking Based on Game-Theory Combination Weighting

Pipe-joint misalignment in pipe-jacking construction is usually confirmed during final quality inspection, but risk signals related to guide-rail gradient deviation, jacking-force utilization, grouting-volume satisfaction, alignment-correction status, and hydrogeological disturbance accumulate during construction. Conventional inspection can determine whether a joint exceeds the allowable limit, but it cannot sufficiently explain risk sources or guide subsequent control. Taking a nominal diameter of 1500 mm (DN1500) steel pipe-jacking project in Wuhan, China, as the study object, this paper establishes a 12-indicator system covering pipe-end geometry and assembly quality, jacking-construction control, and hydrogeological and environmental response. Comprehensive weights are determined by combining a trapezoidal fuzzy analytic hierarchy process (AHP) with an improved criterion importance through inter-criterion correlation (CRITIC) approach using game-theory weighting, after which, a finite-interval cloud model is used to distinguish five risk grades. The results show that the combined weights of water head above the pipe crown, guide-rail gradient deviation, stratigraphic mutation and obstacle score, alignment-correction status, and grouting-volume satisfaction were 0.1803, 0.1737, 0.1702, 0.1027, and 0.0889, respectively, identifying these variables as comparatively influential in this case. All ten evaluated pipe joints were classified as very low risk; however, their memberships in the low-risk and medium-risk grades ranged from 0.0318 to 0.0898 and from 0.0070 to 0.0792, respectively. These differences indicate that the continuous membership vector provides additional information beyond the discrete grade label and can support differentiated post-evaluation rechecking. The proposed framework can therefore provide preliminary support for construction risk early warning and differentiated post-evaluation control, while further validation using larger samples and multiple projects is required.