Online detection of defects in arc additive manufacturing based on infrared temperature field

According to the particularity of layer by layer accumulation in additive manufacturing,the idea of online layer detection could avoid irreparable wastes after finishing processing. A nondestructive detection method was adopted based on real-time acquisition and analysis of the infrared temperature field of the fusion layer. The infrared contour of the fusion layer by stacking multiple frames of temperature field data was obtained,and the forming quality of the current fusion layer was judged according to the extracted infrared contour characteristics. Finally,the infrared contour image was classified and identified through the residual network ResNet18. Through online verification,the accuracy of detection classification could reach 97. 16%,and the detection and classification time of the single infrared contour image was only 1. 35 ms. This method could be used to detect the forming quality in the process of additive manufacturing,which was convenient for subsequent milling,repair welding and optimization of process parameters. It had practical engineering significance.