In order to realize automation of welding

new type of sensor must be developed and filtering and analysis are carried out on collected signal. The working principle of rotating arc sensor is introduced. because of sampled welding current contains a lot of noise

ideal welding current waveform must be obtained through combination of a variety of filtering methods. In order to study welding current waveform

tracking experiments of welding seam which has small bending angle and welding seam which has right angle are carried out. Results show that the welding current waveform after filtering is ideal

through welding current waveform

welding torch can be identified whether it is forward

backward

and leans neither forward nor backward

or welding torch is left

right

or on the middle. Pave the way for deviation identification and control of welding robot.