In this paper, aimed at the problems in arc spectral information in AlMg Alloy pulsed alternating current (AC) GTAW, principal component analysis (PCA) was utilized for redundancy removal and spectral characteristic signals extraction. The spectral wavelengths were classified as the first and the second principal components, associated with Ar I lines and metal lines respectively; pulse interference caused by welding current was eliminated by using mean value during signal’s pulse period (MSP) and wavelet reconstruction. The relations between these extracted signals and different defects were discussed. The results show that the first principal component is closely related to the arc heat input while the second principal component is closely related to the weld pool stability.