Keyhole tungsten inert gas welding（K-TIG） is a new welding technology that can form a keyhole during welding and achieve a large penetration depth. The forming of K-TIG weld is well and the welding quality is high. The K-TIG weld can be obtained with single-sided welding and double-sided shaping

so it has great application prospects in the welding of medium plates. In this paper

the acoustic signal data acquisition software for robot K-TIG welding process was written and the acoustic signal acquisition system was established in order to study the penetration state of K-TIG welding. Real-time acquisition of the acoustic signals in the 304 stainless steel K-TIG welding process enables real-time storage of acoustic signals during K-TIG welding. Finally

the characterization and extraction of time domain signal features are carried out from the aspects of energy

variance

root mean square and kurtosis coefficient.It lays a solid foundation for online melt penetration prediction during K-TIG welding.