Volume Expansion Rate Index Reveals the Damage Process of Surrounding Rock: A Machine Learning-Based Effectiveness Evaluation

Energy sustainability and the establishment of ‘national water network’ are all inseparable from construction underground engineering. Monitoring indices for surrounding rock vital safety management engineering construction, which determines actual state rock. The complexity deep leads to many situations that cannot be predicted explained by existing experience. Therefore, it is necessary identify monitoring index best represents damage. Currently, there no advanced convenient effectiveness evaluation schemes information. To fill technical gap, this study introduces volume expansion rate (VER) proposes a machine learning (ML)-based scheme indices. First, six conditions with different in situ stresses designed, tunnel excavation tests conducted. Second, damage determination experiments using ML classification algorithm performed, accuracy matrix significance scores obtained. results show that: (1) multi-class logistic regression more suitable determining high appropriate outcomes. (2) Under higher stress condition, tangential sensitive (3) As increases, significant demonstrate transition ‘from shallow deep, regional point failure’, describing instability inspiring new criterion