Study on plasma optical signal acquisition and analysis method for <i>in situ</i> process monitoring in laser-TIG hybrid welding

The behavior of laser-TIG hybrid welding plasma is closely related to the welding process. In order to obtain the behavior of plasma more quickly, this paper proposes a new method for in situ monitoring of plasma by acquiring the line spectrum of plasma radiation. By applying narrow-band filtering, the time-varying intensity profiles of specific Mg-I spectral lines (470, 518, and 552 nm) in the plasma radiation generated during the hybrid welding of 5083 aluminum alloy are selectively obtained. These optical signals are then denoised using wavelet packet transform and Savitzky–Golay smoothing filters. Further analysis is performed using empirical mode decomposition, fast Fourier transform, and Hilbert transform to extract comprehensive time, frequency, and time–frequency features. The results demonstrate that, compared to full-spectrum acquisition, the selectively captured line spectra exhibit more pronounced variations in signal intensity and frequency, making them more representative of plasma behavior. Among them, the Mg-I 518 nm line—with the highest intensity—exhibits the strongest sensitivity to changes in plasma dynamics. This targeted spectral approach significantly enhances the accuracy and responsiveness of real-time monitoring. The findings provide a promising strategy for the intelligent in situ monitoring of laser-TIG hybrid welding processes, contributing to improved weld quality and process stability.