主催: 一般社団法人 日本機械学会
会議名: 第37回 計算力学講演会
開催日: 2024/10/18 - 2024/10/20
In recent years, there has been a lot of research into estimating parameters from microstructure data using data assimilation with the phase-field method. In this method, parameters are determined so as to reproduce experimental data of microstructure, but in order to obtain valid information from microstructure data, it is important to carry out a sufficient certainty evaluation of the estimated parameters. In this study, we propose a method for estimating parameters using automatic differentiation, a technique that is widely used in the field of machine learning in recent years, and a method for evaluating the certainty of parameters based on the Hessian matrix calculation of the objective function using automatic differentiation. The proposed method was verified using twin experiments that used pseudo-experimental data generated using simulations, and it was shown to be effective for evaluating certainty of estimated parameter.