TY - JOUR
T1 - Bayesian Approach to Micromechanical Parameter Identification Using Integrated Digital Image Correlation
AU - Gaynutdinova, Liya
AU - Rokoš, Ondřej
AU - Havelka, Jan
AU - Pultarová, Ivana
AU - Zeman, Jan
PY - 2023/3/13
Y1 - 2023/3/13
N2 - Materials with heterogeneous structures exhibit complex physical processes such as delamination, cracks, and plasticity, which require micromechanical parameters for understanding. However, identifying these parameters is challenging due to micro-macro length scale differences, ambiguity in boundary conditions, and required high resolution, among others. While the Integrated Digital Image Correlation (IDIC) method is widely used for this purpose, it suffers from sensitivity to boundary data errors and limited identification of parameters within well-posed problems. To address these issues, a Bayesian approach is proposed for micromechanical parameter identification using the Metropolis--Hastings Algorithm (MHA) to estimate probability distributions of bulk and shear moduli and boundary condition parameters. The proposed approach is compared to two versions of deterministic IDIC method under artificially introduced errors. Although MHA is computationally more expensive and in certain cases less accurate than BE-IDIC, it offers significant advantages, including the ability to optimize a large number of parameters, obtain statistical characterization and insights into individual parameter relationships. The paper highlights the benefits of the non-normalized approach to parameter identification with MHA, which significantly improves the robustness in handling boundary noise, compared to deterministic IDIC. The study considers a fiber-reinforced composite sample under plane strain assumption.
AB - Materials with heterogeneous structures exhibit complex physical processes such as delamination, cracks, and plasticity, which require micromechanical parameters for understanding. However, identifying these parameters is challenging due to micro-macro length scale differences, ambiguity in boundary conditions, and required high resolution, among others. While the Integrated Digital Image Correlation (IDIC) method is widely used for this purpose, it suffers from sensitivity to boundary data errors and limited identification of parameters within well-posed problems. To address these issues, a Bayesian approach is proposed for micromechanical parameter identification using the Metropolis--Hastings Algorithm (MHA) to estimate probability distributions of bulk and shear moduli and boundary condition parameters. The proposed approach is compared to two versions of deterministic IDIC method under artificially introduced errors. Although MHA is computationally more expensive and in certain cases less accurate than BE-IDIC, it offers significant advantages, including the ability to optimize a large number of parameters, obtain statistical characterization and insights into individual parameter relationships. The paper highlights the benefits of the non-normalized approach to parameter identification with MHA, which significantly improves the robustness in handling boundary noise, compared to deterministic IDIC. The study considers a fiber-reinforced composite sample under plane strain assumption.
KW - math.AP
U2 - 10.48550/arXiv.2303.07045
DO - 10.48550/arXiv.2303.07045
M3 - Article
SN - 2331-8422
VL - 2023
JO - arXiv
JF - arXiv
M1 - 2303.07045
ER -