Impact-induced stress wave mechanism-based ASVD-AGSD and its application in high-accuracy feature extraction and localization for thin-walled tubes

Abstract

The thin-walled tail drive shaft is easily damaged by projectile impact in helicopter low-altitude combat. The resulting stress waves exhibit asymmetric, non-periodic, transient, and multimodal aliasing characteristics, which present serious challenges for traditional signal analysis methods in accurate feature extraction and localization. This study proposes a novel method, adaptive singular value decomposition and asymmetric Gaussian sparse decomposition (ASVD-AGSD), which integrates a physical mechanism model with adaptive signal processing. This method constructs an asymmetric Gaussian modulated model (AGMM) based on the impact-induced stress wave mechanism. It adaptively selects SVD embedding dimensions and singular components by composite entropy-guided adaptive particle swarm optimization and signal characteristic analysis, enabling adaptive optimal sparse feature extraction combined with AGMM. Furthermore, a novel impact source localization method is proposed and validated by integrating the PZT-5H piezoelectric sensor array localization optimization model. Simulation and experimental validation showed that the ASVD-AGSD method outperformed mainstream approaches in feature extraction accuracy. The waveform arrival time error is less than 0.78%, with waveform similarity up to 0.99. The extracted first-arriving wave’s major energy characteristic is revealed to be the L(0,2) mode with a mere 0.47% group velocity error. Through analysis of the rise-to-decay time ratio, the radial offset of the projectile is the main physical mechanism leading to the waveform asymmetry. The proposed novel impact localization method achieved a 5.61 mm average localization accuracy under different impact parameters. This study establishes a systematic framework from sensing to localization, providing a reliable real-time impact monitoring and damage assessment new path for thin-walled structures.