Intelligent real-time identification technology of stratum characteristics during slurry TBM tunneling

Abstract

The rapid and accurate identification of geological characteristics during TBM tunneling is a common concern in engineering. A set of real-time dynamic response monitoring and analysis systems was installed in the slurry TBM of a water delivery tunnel in Beijing to analyze the variation of geological characteristics during TBM tunneling. The response mechanism and time–frequency characteristics of slurry TBM acceleration in five strata were deeply studied. The variation trends of thrust, torque and acceleration response characteristic values when strata change were compared, and the results show that the acceleration response is extremely sensitive to strata change. An LSTM network was established that considers slurry TBM driving parameters and acceleration response characteristic values as input parameters and realizes the intelligent identification of five kinds of strata with an accuracy of 97.4% and an F1-Score of 96.38%.