materials characterization

To address the low efficiency and poor robustness of conventional non-convex optimization in constitutive model identification for solid mechanics, this paper introduces a “material fingerprinting” framework. It treats the material’s constitutive response—i.e., standardized experimental measurements (direct or indirect) from uniform or heterogeneous deformation fields—as a unique, discriminative fingerprint; constructs a fingerprint database; and enables rapid model selection via pattern recognition. Crucially, this approach abandons iterative optimization entirely, replacing parameter inversion with a data-driven classification paradigm—thereby ensuring both generality and scalability. Validated on hyperelastic materials, the method achieves high-accuracy identification of candidate models from a single experiment, markedly improving modeling efficiency and noise robustness. It establishes a novel paradigm for intelligent, data-informed material characterization.

This study addresses the challenges of quantifying experimental data informativeness and achieving high data–parameter alignment in constitutive model calibration under complex multiaxial loading. We propose a novel mechanics-informatics paradigm featuring: (i) a first-principles “stress-state entropy” framework to quantify the information content of mechanical experiments; (ii) entropy-guided, high-information-efficiency constitutive learning and specimen design—e.g., high-entropy cruciform and low-entropy Pierrce pure-shear specimens; and (iii) a hybrid Bayesian optimization–transfer learning strategy to mitigate modeling uncertainty. The methodology integrates information theory, anisotropic inelastic constitutive modeling, and experimental mechanics. Experimental validation demonstrates substantial improvements in parameter identification accuracy and robustness; entropy-optimized specimens maximize information yield per test; and transfer learning enables accurate calibration without costly physical experiments, preserving fidelity.

The validity boundary of the J-integral for small-scale fracture mechanics experiments remains ill-defined, hindering reliable fracture parameter extraction at micro- and submicron scales. Method: This study systematically establishes critical criteria for the existence of the HRR (Hutchinson–Rice–Rosengren) asymptotic field underlying the J-integral, quantifying the coupled influence of material yield strength, strain-hardening exponent, and minimum specimen size on the maximum valid J-value (Jₘₐₓ). Combining numerical simulations, semi-analytical fracture analysis, and path-independent J-integral evaluations, we construct the first parametric design map spanning broad ranges of material properties and geometric scales. Contribution/Results: The map enables quantitative mapping between theoretical J-integral validity and experimental feasibility at microscale, providing a universal validity criterion for in situ micromechanical fracture testing. It is further extended to quantitative fracture characterization of hydrogen-embrittled metals, significantly enhancing the reliability and comparability of fracture parameters extracted from small-scale specimens.

To address the challenge of real-time, in-situ monitoring of strain, temperature, and crack propagation in additively manufactured metallic structures operating under harsh conditions—currently reliant on inefficient periodic disassembly—this study pioneers the co-integration of magnetostrictive and thermomagnetic functional materials within microtubes, embedded directly into the metal matrix during additive manufacturing. Leveraging electromagnetic coil impedance modulation and eddy-current non-destructive evaluation principles, the approach enables wireless, passive, multi-parameter sensing within the structural interior. The method detects plasticity onset and fatigue crack initiation/propagation thousands of cycles earlier than conventional techniques, enabling condition-based maintenance. Experimental validation demonstrates strain measurement accuracy of ±27 με (full-scale 600 με), temperature accuracy of ±0.75 °C (0–70 °C, 95% confidence level), significantly enhancing in-service structural health awareness and service-life prediction fidelity.

This study addresses the challenge of integrating and interpreting heterogeneous, multiscale characterization data of carbon nanotube (CNT) thin films arising from microstructural variations. To this end, the authors propose an interpretable multimodal machine learning framework—the first application of multimodal explainable AI to multiscale materials characterization. The approach fuses multimodal data, including morphological features extracted from SEM images via binarization, skeletonization, and network analysis, along with Raman spectroscopy, specific surface area, and surface resistivity measurements. By combining UMAP for dimensionality reduction and XGBoost for nonlinear regression, the framework enables end-to-end performance prediction. Under leave-one-out cross-validation, the model achieves optimal predictive accuracy. Feature importance analysis reveals that surface resistivity is primarily governed by inter-junction transport length, crystallinity, and network connectivity, while specific surface area is dominated by junction density and pore size, thereby elucidating the microstructure–property relationships in CNT films.

This study addresses the challenge of accurately identifying impact parameters—velocity, mass, and energy—in aerospace composites under data degradation or noise interference. To this end, the authors propose a unified modeling framework that synergistically integrates physical priors with data-driven learning. By constructing an input space based on a physics-informed energy metric, designing a decoupled surrogate model, and incorporating a hybrid physics-constrained loss function within a neural network architecture, the approach systematically embeds observational, inductive, and learning-based physical biases. This enables decoupled inference of impact parameters while ensuring kinetic energy consistency. Experimental results demonstrate that the method achieves mean absolute percentage errors below 8% for both impact velocity and mass, and below 10% for energy, exhibiting robust generalization and stability even under data sparsity, high noise levels, and damage-induced conditions.

This study addresses the limitations of traditional constitutive models, which oversimplify material behavior, and purely data-driven approaches, which often lack physical consistency, in accurately predicting the stress–strain response of additively manufactured materials. To overcome these challenges, the authors propose a segmented physics-informed machine learning (PIML) framework that leverages process parameters to predict yield points and separately models the elastic and plastic deformation stages. The framework innovatively embeds Hooke’s law, the Voce hardening law, and the Hollomon equation into an LSTM architecture through both loss-function constraints and activation functions. Experimental validation on four additively manufactured materials demonstrates that the activation-based PIML model achieves superior performance, with an average mean absolute percentage error (MAPE) of 10.46 ± 0.81% and an R² of 0.82 ± 0.05, significantly outperforming conventional constitutive models and pure data-driven methods.

Traditional constitutive modeling struggles to integrate multimodal, multifidelity experimental data and faces a trade-off between physical consistency and discovery efficiency. This work proposes paFEMU, a novel framework that, for the first time, combines sparse regression–driven constitutive discovery with finite element model updating. By leveraging only a few simple mechanical tests and full-field digital image correlation (DIC) data, the method employs physics-informed neural networks and adjoint-based optimization to efficiently identify interpretable constitutive relationships. The resulting models are low-dimensional, physically consistent, and amenable to transfer learning across materials. Moreover, they can be seamlessly integrated into existing finite element simulation workflows, offering a practical pathway toward data-driven, yet physically grounded, material modeling.

This work addresses the limited reproducibility and data reusability in atomistic simulations, which often stem from fragmented scripting practices, inconsistent metadata, and inadequate provenance tracking. To overcome these challenges, we propose the first reusable simulation framework that integrates semantic workflows with a materials application ontology. By structuring metadata for key mechanical and thermodynamic properties, our approach enables automated provenance capture and generates outputs compliant with FAIR (Findable, Accessible, Interoperable, Reusable) principles. The framework supports workflow reuse across different interatomic potentials and material systems, successfully validating structure–property relationships such as the Hall–Petch effect. It produces standardized, interoperable, high-quality datasets, establishing a knowledge-driven paradigm for AI-ready atomistic simulations.

This study addresses the limitations of conventional approaches in characterizing the elastoplastic mechanical behavior of heterogeneous materials such as welded joints, which often suffer from insufficient spatial resolution or low reliability. The authors propose an extended virtual fields method that enables high-resolution reconstruction of spatially varying constitutive parameters without requiring prior knowledge of their distribution. By integrating optical measurements, synthetic equivalent data, and numerical validation, the method is shown to be robust across diverse weld geometries, loading conditions, and dissimilar material combinations. Numerical experiments demonstrate that the proposed approach accurately converges to the target parameter fields, achieving precise spatial mapping of heterogeneous mechanical properties and offering a novel pathway for the mechanical characterization of complex welded structures.