Voiceprint analysis for non-destructive monitoring of fastener preload variations in complex structural components

• Proposes Voiceprint Timbre Difference Analysis (VTDA), a quantitative method for detecting preload force variations, outperforming conventional acoustic non-destructive testing. • Achieves 98.2% accuracy in fastener failure detection, validating high reliability and robustness. • Integrates ancient acoustics with deep learning, bridging materials science, mechanical engineering, and archaeological conservation, with broad applications in industrial and cultural diagnostics. Voiceprint technology is primarily utilized for individual feature recognition. However, in engineering contexts, accurately and rapidly detecting hidden damage and subtle changes in complex structural components, particularly internal invisible damages, presents a significant challenge for conventional voiceprint recognition methods. In this study, we introduce a non-intrusive auditory evaluation method, Voiceprint Timbre Difference Analysis (VTDA), to assess the consistency of sound transmission performance in complex structural solids. This method treats the tested object as a musical instrument and can be performed online or on-site without the need for specialized testing environments. The study proposes a non-destructive structural health monitoring (SHM) method through voiceprint timbre analysis, and introduces a robust sound classification algorithm developed using PaddlePaddle, an open-source artificial intelligence platform. The algorithm is suitable for training both voiceprint recognition and sound classification models to accurately identify the locations of loosened screws. Importantly, VTDA enables precise localization of metal fasteners at specific positions on stressed members that have experienced impact. This work develops a scalable, physics-informed approach to predictive maintenance, which holds transformative potential for rail, aerospace, and civil engineering applications where real-time integrity monitoring is critically important.