Analysis of Hygrothermal Aging Mechanism of Composites from Acoustic Emission Data Using Machine Learning and LLMs

Acoustic Emission (AE) technology is a key technical means for real-time damage monitoring of composite materials.However, its accuracy in identifying composite damage mechanisms and capacity for deep information mining remain constrained by the traditional paradigm of analyzing individual features of high-dimensional AE signals one by one. Addressing the demand for analyzing the hygrothermal aging damage mechanisms of fiber-reinforced composites, this study employs artificial intelligence (AI) for information mining of AE signals synchronously collected during three-point bending tests. With the core technical chain of "feature correlation analysis, model discrimination, SHAP interpretability analysis, data dimensionality reduction, Large Language Model (LLM)", the visualization and low threshold of the evolution of hygrothermal aging damage mechanisms in composite materials are achieved. This research provides a general technical paradigm for the visualization of damage mechanisms in various service scenarios of composite materials.