Parametric approach to positional fault detection in short arc welding

Abstract This work investigates a parametric technique for short arc welding fault detection. The independent component analysis algorithm is applied to split welding voltage and current signals into arcing and short circuiting regions. A series of parameters including means, medians and normalised inner products are calculated by processing the different sections of welding electrical data. Out of position welds in overlap joints are studied with investigation of the connection between parameters and short arc welding features. The effect of changing shielding gas is also illustrated. Figures of merit to quantify the change in various parameters between faulty and good welds are compared and discussed. It is shown that the data separation into two regions with independent component analysis is useful in providing various parameters that can assist in short arc welding fault detection under different conditions.