During disk laser welding bead-on-plate welding of type 304 austenitic stainless steel plates

a near infrared band highspeed camera

an ultraviolet band high-speed camera and a visible light band high-speed camera are used to capture the dynamic images of the keyhole

plume and spatters in the laser welding at the same time. The keyhole

plume and spatters are defined as the characteristic parameters of welding process

and the dynamic characteristics are considered as the output parameters of the neural network

and the weld width is considered as a parameter reflecting the welding status and the input parameter of the neural network. A technique that uses multifeature information fusion of keyhole

plume and spatters to obtain the welding status of high-power disk laser welding based on neural network technology is presented. The defined keyhole

plume and spatters characteristic parameters are considered as inputs to calculate theoretical weld width. The proposed BP neural network has been successfully tested in a real time disk laser welding process.