Joint analysis of acoustic emission data using machine learning and LLM reveals the mechanism of wet heat aging in composite materials

Research Overview

A study published in Polymer Testing, Volume 163, October 2026, by Wenhan Tian, Yichuang Zhang, Han Wang, Yan Gao, Hongfei Zhou, and others, titled “Analysis of hydrogeological aging mechanisms of composites from acoustic emission data using machine learning and LLMs”. This study focuses on the aging mechanism of composite materials in humid and hot environments, using acoustic emission (AE) data as a monitoring method and introducing machine learning and large language models (LLMs) for data analysis.

Methods and Technical Paths

Research is based on acoustic emission data, combined with machine learning algorithms and LLMs to analyze the damage signals during the wet heat aging process of composite materials.

. Acoustic emission technology can capture transient elastic waves generated during the aging process of materials, which are related to damage modes such as matrix cracking, fiber fracture, and interface debonding. By using machine learning for pattern recognition of acoustic emission features and using LLMs to assist in interpreting the analysis results, this study attempts to establish a correlation path from monitoring data to aging mechanisms.

Significance in the field of polymer aging and stabilization

Although this study did not directly involve the formulation of polymer additives, its analytical method of aging mechanism has reference value for polymer aging and stabilization technology.

. In weather resistant and stable systems, the humid and hot environment is one of the important factors that trigger polymer degradation. Understanding the damage evolution during aging can help evaluate the actual effectiveness of antioxidant and photo stable systems. Acoustic emission, as a non-destructive testing method, can complement traditional aging characterization methods and provide more dimensional data support for the stability evaluation of HALS, UVA, and composite systems in complex environments. In addition, the application of machine learning and LLMs in aging data analysis suggests that polymer aging detection and characterization are developing towards data-driven and intelligent directions. For engineering and technical personnel engaged in research on processing stability and weather resistance stability, such methods are expected to improve the efficiency of aging mechanism research and provide richer experimental basis for formula optimization of composite systems.

Research on boundaries and observation directions

It should be pointed out that the object of this study is composite materials, and their acoustic emission signal characteristics may differ from those of pure polymer systems.

. The study did not involve specific antioxidant systems, HALS or UVA product formulation data, nor did it provide a quantitative relationship between the dosage of additives and aging performance. Therefore, when transferring this method to the evaluation of polymer additives, it still needs to be validated in conjunction with specific material systems. It is worth noting whether such data-driven methods can form a standardized process for early warning of polymer aging and screening of stable systems.

Information source:Analysis of hygrothermal aging mechanism of composites from acoustic emission data using machine learning and LLMs， Organized by Herun Chemical.