A risk assessment approach for road collapse along tunnels based on an improved entropy weight method and K-means cluster algorithm

Ground collapse often occurs when strata are disturbed, particularly during tunnel construction, threatening the safety of people and underground facilities. Therefore, evaluating the risk of road collapse during and after tunnel construction is essential. Considering this, an improved entropy weight method was adopted to establish an evaluation system for assessing the collapse risk level of the Qianjin West Road, through which the Kunshan metro line S1 passes. The main implementation steps of the evaluation system include ten assessment indices, data visualization of indices, weight determination for indices, a cluster algorithm for the classification of results, and analysis of the evaluation results. In contrast to the traditional single method, and to improve the reliability of the risk assessment results, the method proposed in this paper considers two aspects. First, to eliminate subjective or objective biases generated by a single evaluation method, the entropy weight method (EM) and analytic hierarchy process (AHP) were integrated to formulate the improved entropy weight method. Second, to divide the evaluation results more reasonably, the k-means cluster algorithm using the Calinski Harabasz method was applied. The reliability and accuracy of the proposed improved entropy weight method are verified based on the case of Line S1. Furthermore, a method comparison and the concept of the coefficient of variation method were adopted to demonstrate the reliability of the risk assessment results. The improved entropy weight method can be used to evaluate ground collapse risk levels during and after tunnel construction.