Acoustic-based arc stability assessment in MAG-WAAM using transform-domain signal features

Reliable in-process indicators of arc instability are needed for wire arc additive manufacturing (WAAM), because conventional post-process inspection cannot support timely corrective action. This study evaluated whether compact transform-domain descriptors of airborne arc sound can distinguish pronounced deposition irregularities during first-layer MAG-WAAM. A full-factorial matrix of 36 single-bead trials was produced using 1.2-mm ER70S-6 wire by varying arc voltage (20–32 V), wire-feed rate (6–10 m/min), and travel speed (145–435 mm/min). Five-second acoustic records were acquired at 10 kHz. The standard deviations of selected FFT, DCT, and Level-2 Haar-DWT coefficient sets were calculated and combined multiplicatively into an empirical index, M; total and mean-square signal energy were evaluated as complementary amplitude-based descriptors. Four trials were operationally classified as unstable from deposition continuity, bead appearance, and visible spatter. These trials yielded M = 0.0165–0.0249 and E = 93.38–108.35 a. u., whereas the remaining trials yielded M = 0.0006–0.0054 and E = 7.93–30.86 a.u. Accordingly, the empirical limits M >0.01 and E >80 a. u. Provided complete descriptive separation within this dataset. Exploratory log-linear models were significant for ln(E) ( R 2 = 0.5405 ) and ln(M) ( R 2 = 0.6313 ) ; wire-feed rate showed a significant positive association, voltage a significant negative association, and travel speed no significant independent association. The results establish proof-of-concept acoustic discrimination under a fixed acquisition configuration, but the thresholds require independent calibration and validation before multilayer or cross-system application.