arXiv subjects

Jinghui Li

Publications and source records attributed to Jinghui Li.

5 recordsLinked to original sources

arXiv2026

Measurement of multiple mechanical properties from multi-dimensional signals in nanosecond laser ablation via PINN

Accurate evaluation of mechanical properties in steels under ageing or service conditions remains a major challenge. We propose a thermo-mechanical coupling framework for nanosecond laser ablation based on energy conservation, which is embedded into a physics-informed neural network (PINN) to enable simultaneous inversion of multiple mechanical properties. A thermo-mechanical coupling coefficient is defined to uniformly describe the dynamic allocation of input laser energy among thermal diffusion, mechanical work and plasma shielding across different deformation stages under laser irradiation. Furthermore, hard-to-measure physical characteristics in the coupled equation are replaced with experimentally accessible features obtained through the simultaneous acquisition of spectroscopic, shockwave and surface-wave signals. Using 210 experimental datasets, the framework simultaneously recovers Young's modulus, yield strength, ultimate tensile strength and micro-Vickers hardness with high accuracy (R2=0.9927, 0.9912, 0.9916 and 0.9959 respectively), significantly outperforming the baseline method (ultrasonic velocity regression for E, R2=0.0012). Comparisons with linear normalization and unconstrained neural networks demonstrate that PINN achieves near-unity accuracy through the embedding of conservation-law constraints. Partial dependency analysis further uncovers the nonlinear coupling laws between input features and mechanical properties. The proposed paradigm, integrating conservation laws, measurable features and physics-informed learning, offers a universal approach for non-contact, high-precision and physically consistent multi-to-multi inversion of multiple material properties under nanosecond laser ablation conditions.

physics.plasm-ph↗

arXiv2026

Physics-informed genetic algorithms (PIGAs) facilitating LIBS spectral normalization with shockwave characteristics

Inspired by physics-informed neural networks (PINNs) inheriting both the interpretability of physical laws and the efficient integration capability of machine learning, we propose a framework based on stoichiometric ablation for LIBS spectral normalization, encoding physical constraints between LIBS intensities and shockwave characteristics (temperature Tshock and pressure P) into optimization algorithms with multiple independent objectives, named physics-informed genetic algorithms (PIGAs). It is characterized by its applicability to the wider laser energy range covering laser-induced breakdown to significant plasma shielding and spectral lines undergoing self-absorption outperforming the widely-used physical linear or multivariate data-driven normalization methods. The home-made end-to-end LAP-RTE codes serves as the benchmark to validate the physical reciprocal-logarithmic transformation and its extensibility to self-absorption spectral lines for PIGAs. Next experimental spectral lines are statistically used to validate PIGAs correction effects, the median RSDs of spectral intensities can be effectively reduced by 85% (corrected by P) and 88% (corrected by Tshock) for 108 Fe I lines, while for 33 Fe II lines, reduced by 77% (corrected by P) and 86% (corrected by Tshock). Seventeen self-absorption lines are also corrected effectively, with RSDs being reduced by 78% (corrected by P) and 89% (corrected by Tshock). Our proposed idea of combining optimization methods to quantify unknown parameters in normalization strategies can also be extended to excavate the correlation between parameters for other low-temperature plasma fields with similar processes.

physics.plasm-ph↗

arXiv2026

Coupling model of metallic target ablation-plasma evolution-radiation under nanosecond laser irradiation

The interaction of nanosecond laser pulses with metallic materials involves multiple complex physical processes. It is challenging to construct a self-consistent model capable of uniformly describing all stages. This work establishes a multi-physics coupling model for pure iron, encompassing laser energy deposition, solid-liquid phase transition, gas-liquid interfacial kinetic transport, plasma expansion and ionization, and spectral radiation. The numerical solution adopts a partition method, utilizing an implicit compact difference scheme for the target and a Mac-Cormack explicit scheme for the plasma. The simulations elucidate the emergence of plasma shielding and its inhibitory effect on the evaporation process, thereby confirming that 81.6% of the early-stage ablation products are transported through a supersonic expansion mode. The model successfully captures the complete evolution of the plasma plume from a high-temperature, highly ionized state to a low-temperature, neutral atomic state. Based on this, spectral calculations demonstrate the dynamic evolution of radiative characteristics from an early stage featuring a strong continuum background dominated by ion lines to a later stage where the continuum attenuates, atomic lines become prominent, and self-absorption appears. The emergence of self-absorption proves the ability of the model to effectively capture the optical thickness effects arising from spatial inhomogeneity within the plasma. Through systematic comparison between experimentally measured spectra and calculated results from the PrismSPECT and NIST LIBS spectral programs, the model presented here achieves the highest comprehensive scores in quantitative evaluations of multiple channels. This validates the necessity and superiority of the full-chain self-consistent modeling approach, especially in describing plasma inhomogeneity and radiation transport.

physics.plasm-ph↗

arXiv2026

Evaluating LLMs for Demographic-Targeted Social Bias Detection: A Comprehensive Benchmark Study

Large-scale web-scraped text corpora used to train general-purpose AI models often contain harmful demographic-targeted social biases, creating a regulatory need for data auditing and developing scalable bias-detection methods. Although prior work has investigated biases in text datasets and related detection methods, these studies remain narrow in scope. They typically focus on a single content type (e.g., hate speech), cover limited demographic axes, overlook biases affecting multiple demographics simultaneously, and analyze limited techniques. Consequently, practitioners lack a holistic understanding of the strengths and limitations of recent large language models (LLMs) for automated bias detection. In this study, we conduct a comprehensive benchmark study on English texts to assess the ability of LLMs in detecting demographic-targeted social biases. To align with regulatory requirements, we frame bias detection as a multi-label task of detecting targeted identities using a demographic-focused taxonomy. We then systematically evaluate models across scales and techniques, including prompting, in-context learning, and fine-tuning. Using twelve datasets spanning diverse content types and demographics, our study demonstrates the promise of fine-tuned smaller models for scalable detection. However, our analyses also expose persistent gaps across demographic axes and multi-demographic targeted biases, underscoring the need for more effective and scalable detection frameworks.

cs.CL↗

arXiv2025

Diverse Human Value Alignment for Large Language Models via Ethical Reasoning

Ensuring that Large Language Models (LLMs) align with the diverse and evolving human values across different regions and cultures remains a critical challenge in AI ethics. Current alignment approaches often yield superficial conformity rather than genuine ethical understanding, failing to address the complex, context-dependent nature of human values. In this paper, we propose a novel ethical reasoning paradigm for LLMs inspired by well-established ethical decision-making models, aiming at enhancing diverse human value alignment through deliberative ethical reasoning. Our framework consists of a structured five-step process, including contextual fact gathering, hierarchical social norm identification, option generation, multiple-lens ethical impact analysis, and reflection. This theory-grounded approach guides LLMs through an interpretable reasoning process that enhances their ability to understand regional specificities and perform nuanced ethical analysis, which can be implemented with either prompt engineering or supervised fine-tuning methods. We perform evaluations on the SafeWorld benchmark that specially designed for regional value alignment. Experimental results demonstrate our framework significantly improves LLM alignment with diverse human values compared to baseline methods, enabling more accurate social norm identification and more culturally appropriate reasoning. Our work provides a concrete pathway toward developing LLMs that align more effectively with the multifaceted values of global societies through interdisciplinary research.

cs.AI↗