[17a-S2_204-1~13] 23.1 Joint Session N "Informatics"

9:15 AM - 9:30 AM JST(12:15 AM - 12:30 AM UTC)

[17a-S2_204-2] Constitutive Equation Identification for Complex Fluids by Sparse Regression with Material Objectivity

〇(B)Shotaro Moro¹, Takeshi Sato², Shota Kato¹, Katsuaki Tanabe¹, Souta Miyamoto¹ (1.Kyoto Univ., 2.Kanazawa Univ.)

Keywords:

Sparse Regression,Material Objectivity,Constitutive Equation

To predict the flow behavior of complex fluids such as polymer melts, we propose a data-driven method for identifying constitutive equations that satisfy material objectivity. In this approach, we construct a library of candidate terms that are invariant under frame transformations and perform sparse identification using the SINDy algorithm. We apply the method to time-series data obtained from Brownian dynamics simulations under oscillatory shear and identify the deviation from the Upper-Convected Maxwell (UCM) model. Predictions based on the identified equation reproduce the simulation results well under interpolative conditions, whereas the generalization performance under extrapolative conditions remains a challenge.