Inversion method of initial seepage field based on reduced order projection
In order to solve the time-consuming problem of traditional parameter inversion based on reduced order models,this study proposes an improved reduced order inversion method(IterROM)for performing fast parameter inversion tasks.Unlike the traditional reduced order inversion method,which trains first and then inverts,this method tightly integrates pa-rameter inversion with reduced order model training,that is,using the reduced order model in training to replace the original model for parameter inversion,and applying the inversion results to the training of the reduced order model.This method u-ses observation data and a posterior error to form the objective function during the parameter inversion stage,and uses the maximum node error between the reduced order model solution and the full order model solution as the iteration termination accuracy.After testing the initial seepage field inversion model of a certain project,it was found that IterROM has higher ac-curacy and lower computational cost compared with traditional methods,and it can be used to quickly perform the inversion task of the initial seepage field.
initial seepage fieldreduced order modelparameter inversionoptimization algorithmpermeability coefficient
国家科技期刊平台
NSTL
万方数据
