Effectiveness Evaluation of Surrounding Rock Monitoring Index Based on Machine Learning
Underground engineering is advancing into deeper strata,where phenomenon beyond empirical rec-ognition frequently occur in practice.There is a need to effectively assess the diagnostic capability of surrounding rock monitoring indicators for the damage state of the surrounding rock.The volume expansion rate(VER)in-dex of surrounding rock is introduced,and a method for evaluating the effectiveness of surrounding rock monitor-ing indexes based on machine learning is proposed.The data samples are obtained from the discrete element test.Six conditions of ground stress levels are designed.The displacement,stress and rock mass damage are monitored in real time after tunnel excavation.Secondly,experiments for surrounding rock damage determina-tion and indicator assessment are carried out using machine learning techniques.The results show that:The sur-rounding rock monitoring data obtained from the discrete element test are reasonable and effective;The damage determination results based on classification algorithm have high accuracy;With the increase of in-situ stress,the significant monitoring indexes show a change process from shallow to deep,from regional damage to point failure.Under the condition of higher in-situ stress level,tangential stress is more sensitive to the change of sur-rounding rock state.This method innovatively evaluates the effectiveness of surrounding rock monitoring indica-tors and provides a new criterion for monitoring and assessing the instability of surrounding rock.
surrounding rock damage judgmentdiscrete element methodmachine learningvolume expansion ratesurrounding rock instability
NETL
NSTL
万方数据
