Deviation recognition of high speed rotational arc sensor based on support vector machine

Abstract

Signal patterns of high speed rotational arc sensor in gas metal arc welding (GMAW) have been studied. For V-groove butt joint, a geometry model of the weld bead profile and torch rotating has been developed. Welding current waveforms of both simulations and experiments have been analyzed. The welding current waveforms simulated based on the mathematical model are consistent with those captured in welding experiments, which proves that the mathematical model is correct. The signal features are analyzed as torch deviation from V-groove centre varied. The results show that the deviation of the welding torch is in proportion with the asymmetry of the current waveform in corresponding arc rotational cycle. A SVM is used to recognize the torch deviation. The results of this study are helpful to the design and application of high speed rotational arc sensors.