Probabilistic Predictions of the Convergences of Surrounding Rock Masses in Underground Rock Caverns

Abstract

The convergences of surrounding rock masses are important indexes which can be used to evaluate the stability levels of underground rock excavations. The predictions and control of convergences are essential tasks to ensure that safety measures are met in such rock excavation processes as mining, tunneling, and underground constructions. Meanwhile, many uncertainties are associated with the predictions of convergences due to the complexity and nonlinearity of rock mechanical behaviors. However, such uncertainties had not been considered in previous studies. In this paper, a novel method was proposed to predict the convergences and uncertainties of surrounding rock masses in underground caverns. In the present study, a relevance vector machine (RVM) was adopted to build the relationships between the convergence displacement and mechanical parameters of the rock masses. Numerical simulations and experimental designs were used to provide the samples for the RVM model. Then, Monte Carlo simulation (MCS) was utilized to simulate the uncertainties of the convergences based on the RVM model. The proposed method was verified using different examples with the simulated uncertainties. The results showed that the proposed method had the ability to reasonably predict the convergences and their related uncertainties. When compared with the deterministic method, the proposed method was found to be more rational and scientific and had also conformed to rock engineering practices. Therefore, the proposed method provided a scientific way to predict convergences and displacement uncertainties of surrounding rock masses in underground caverns.