Method of Acoustic Emission Data Quality Assessment and Fatigue Damage Monitoring for Aeronautical Structures

To address the challenges in monitoring micro-crack initiation and propagation in aeronautical structures using acoustic emission (AE) technology, including difficulties in extracting micro-damage features and low damage localization accuracy under noise interference, this study proposes a method that enhances aeronautical structural health monitoring through AE data quality assessment and fatigue damage monitoring for aeronautical structures. The method establishes an intelligent quality assessment model for AE signals based on a deep convolutional autoencoder, which extracts high-level features from raw AE data to achieve adaptive discrimination between crack-induced signals and noise, enabling automatic denoising of AE signals. Experimental validation is conducted using AE monitoring data collected from fatigue tests of aerospace aluminum alloy components. The results demonstrate that the proposed method yields average reconstruction errors of 0.007 and 0.020 for data from the early healthy stage and mid-to-late damage stage, respectively, achieving accurate differentiation between noise and damage signals. The detected damage initiation time is 22 min earlier than the macroscopic crack observation time in the experiment, proving effective for early warning during crack initiation and propagation. Compared to original localization maps, the damage progression localization after noise removal clearly reveals the long-term crack development trend. The experimental results demonstrate the potential applicability of the proposed method in engineering scenarios.