Unascertained Measure-Set Pair Analysis Model of Collapse Risk Evaluation in Mountain Tunnels and Its Engineering Application

Abstract

Collapse is a major geological disaster that occurs during mountain tunnel construction. However, the accuracy of collapse risk assessment is generally restricted by evaluation factors or methods. This paper proposes a novel integrated collapse risk evaluation method for mountain tunnels based on case-based reasoning, rough set theory, and unascertained measure-set pair analysis (UM-SPA) theory. First, the risk surroundings and risk factors involved in tunnel collapse are summarized by the analytic hierarchy process, and a preliminary risk evaluation index system is established. Then, an attribute reduction algorithm based on the conditional information entropy of a rough set is proposed and applied to cases with similar attribute characteristics as those of the tunnel to be evaluated, to remove the relatively insignificant or redundant indices and improve the reliability of risk assessment. Finally, taking the relationship between tunnel collapse and its evaluation indices as an unascertained system, Set Pair Analysis (SPA) theory is introduced to optimize the credible degree recognition criteria of unascertained measure theory (UMT). Combined with the modified entropy weight method, a UM-SPA model for tunnel collapse risk evaluation is established to calculate the level of collapse risk quantitatively and predict the development trend of risk dynamically. Taking Xiucun Tunnel passing through fault F18 as an example, collapse risk is evaluated and compared with the evaluation results of traditional UMT and field status. The results demonstrate the feasibility and efficiency of the proposed approach and provide a new idea for collapse risk prediction while constructing mountain tunnels.