A new back-analysis method based on ν-SVR and improved PSO algorithm and its application

Challenged by diversity,complexity and precision in geotechnical engineering practices,back-analysis methods are required to quickly obtain feedback parameters for numerical simulations on the basis of monitoring data with fewer but more elaborate forward numerical simulations.Thanks to the specialties of support vector regression machine(ν-SVR) and improved partical swarm optimization(PSO) algorithm with variable neighborhood,a method and process for geotechnical back-analysis is set up.And to prove the correctness and validity of the proposed method,a case study of back-analysis of the left slope of Jinping-Ⅰ hydropower station is carried out.According to the monitoring data of troublesome profile Ⅱ1-Ⅱ1 in project site,critical deformation parameters for forward numerical simulations are fed back with the proposed method,and the results of further simulation with the feedback parameters match the monitoring data fairly well.