Deformation Forecasting for Tunnel Rock by Gaussian Process Machine Learning Model
Qimao Liu · Journal of Guilin University of Technology · 2010
As deformation of surrounding rock mass is a highly complicated nonlinear time series problem,GP models based on static and dynamic for deformation forecasting are proposed.Based on historical deformation and establishing deformation of surrounding rock mass time series,the nonlinear relationship between present deformation of surrounding rock mass and historical deformation of surrounding rock mass is built by GP model.The case study shows that the GP machine learning model forecasting deformation of surrounding rock mass of tunnel is feasible.Without establishing complicated model of rock mass mechanics,the GP model can forecast nonlinear deformation of surrounding rock mass of tunnel reasonably according the historic measured information of deformation.The forecasting result based on static knowledge base is superior to that based on dynamic knowledge base.