Quality Detection of Laser Welding Based on One-Dimensional Convolutional Neural Network

During the laser welding process, the occurrence of abnormal conditions can cause workpiece defects and make the products fail. Therefore, a real-time monitoring system required to detect the laser welding process for ensuring the product quality. In this paper, we develop a real-time laser welding data acquisition equipment that can measure the plasma density, laser light intensity, and melt pool temperature. In addition, a convolutional neural network is built to achieve rapid detection of laser welding quality. The experimental results show that the proposed method can quickly detect the quality of the laser welding process, and the average accuracy for the welding defects has reached 97%.