A novel real-time monitoring approach for surface integrity in laser shock peening: spectro-acoustic data fusion

A spectro-acoustic data fusion approach was employed for real-time monitoring of surface integrity during the laser shock peening (LSP) process, revealing the mechanism of plasma physical field interaction. The apparent plasma temperature and electron density were quantified through spectral analysis, while the propagation behavior of the shock wave was analyzed using the time–frequency characteristics of the airborne acoustic signal. A multivariate regression model was developed using the spectral intensity ratio of Hα/N II 500.515 nm and the peak-to-peak amplitude of the acoustic wave. This model achieved high-precision prediction of surface residual stress (R² = 0.944) and Vickers hardness (R² = 0.946) for γ-TiAl alloy, improving by over 5 % compared to single-signal regression models. A spectro-acoustic fusion diagnostic model incorporating a cross-modal attention mechanism within the Transformer architecture was established, achieving a prediction accuracy of 99.52 %. This study provides significant theoretical and practical contributions to the adaptive LSP of critical aerospace components.