AI Meets Plasticity: A Comprehensive Survey

TL;DR

This comprehensive survey diagnoses how artificial intelligence intersects with materials plasticity, spanning microscopic mechanisms (dislocations, grain structure) to macroscopic constitutive behavior. It organizes AI methods from classical ML to deep learning, physics-aware neural networks, probabilistic UQ, and generative AI, framing applications as surrogate models, microstructure characterization, and multiscale analysis. Key contributions include a unified taxonomy of AI methodologies, a review of data sources and sampling strategies, and practical guidance on evaluation metrics, best practices, and future directions for physics-informed and agentic AI in plasticity. The work highlights the potential of DL, GNNs, and neural operators to capture history dependence, microstructural effects, and multiscale couplings, while underscoring the need for physical constraints and uncertainty quantification to enable reliable, scalable deployment in materials design and simulation.

Abstract

Artificial intelligence (AI) is rapidly emerging as a new paradigm of scientific discovery, namely data-driven science, across nearly all scientific disciplines. In materials science and engineering, AI has already begun to exert a transformative influence, making it both timely and necessary to examine its interaction with materials plasticity. In this study, we present a holistic survey of the convergence between AI and plasticity, highlighting state-of-the-art AI methodologies employed to discover, construct surrogate models for, and emulate the plastic behavior of materials. From a materials science perspective, we examine cause-and-effect relationships governing plastic deformation, including microstructural characterization and macroscopic responses described through plasticity constitutive models. From the perspective of AI methodology, we review a broad spectrum of applied approaches, ranging from frequentist techniques such as classical machine learning (ML), deep learning (DL), and physics-informed models to probabilistic frameworks that incorporate uncertainty quantification and generative AI methods. These data-driven approaches are discussed in the context of materials characterization and plasticity-related applications. The primary objective of this survey is to develop a comprehensive and well-organized taxonomy grounded in AI methodologies, with particular emphasis on distinguishing critical aspects of these techniques, including model architectures, data requirements, and predictive performance within the specific domain of materials plasticity. By doing so, this work aims to provide a clear road map for researchers and practitioners in the materials community, while offering deeper physical insight and intuition into the role of AI in advancing materials plasticity and characterization, an area of growing importance in the emerging AI-driven era.

Figures (32)

• Figure 1: AI meets plasticity: when data-driven approaches intersect with physics-driven approaches, new paradigms emerge.
• Figure 2: Illustration showing the diverse data sources in material science.
• Figure 3: (a) Schematic representation of a SVM. (b) (Left) Yield surface of diamond in 3D stress space dyckhoff2023data, (Right) Discrepancies between the phenomenological model and experimental observations can occur when the model fails to capture inherent anisotropy or asymmetry in the material behavior. In such cases, the phenomenological framework is locally refined using a SVM component to accurately reconstruct the reference yield surface at the measured data points fuhg2023enhancing. (c) (Left) Schematic of a single-output SVM architecture, (Right) Schematic of a multi-output SVM architecture for hypoplastic geomaterials zhao2015material.
• Figure 4: Schematic illustrations of (a) a typical DT architecture and its terminology, (b) a bagging architecture as a parallel ensemble method, and (c) a boosting architecture as a sequential ensemble DT method.
• Figure 5: (a) Schematic of a typical DT model with temperature, strain and strain rate as inputs lim2020flow. (b) Schematic of a RF model with three inputs and and an output indicating the aggregation of each tree result poluru2025constitutive.