【摘要】 本文尝试对焊接过程进行多信号同步采集。在计算机视觉库OpenCV处理的熔池表面图像基础上,多焊接信号数据结合建立了基于机器学习的焊缝熔深预测模型,从而得到了焊接熔池真实形貌。利用双椭球焊接热源模型计算焊接温度场,并提出了外接立方体和抽样方法,完成了多信号预测熔池对数值模拟熔池的修正试验研究。有限元模拟熔池和多信号预测熔池长度上和宽度上的平均误差分别为2.77%和6.55%,焊缝熔深的平均误差为8.74%。基于多信号测试得到的焊接熔池,为实际焊接过程中的温度场数值模拟提供了热源模型校准模型。更多还原

【Abstract】 Studying the welding temperature field in the welding process is very important. In this paper,we are to measure the actual molten pool in the welding process to modify a simulated temperature field. It can be realized by synchronous acquisition of different welding process information by various sensors.Computer vision library OpenCV is used to process the image of molten pool and obtain the characteristic information of the molten pool image. According to these data and information,the weld penetration is predicted by machine learning. At the same time,a numerical simulation of welding process is carried out and the temperature field of welding process is successfully simulated. A method of external cube and sampling is proposed to modify the simulated molten pool by actual molten pool images. Eventually,good results are obtained. The average error of the length and the width of simulated and actual molten pool are2. 77% and 6. 55% respectively and the average error of weld depth is 8. 74%. The experimental study of temperature field model based on an actual molten pool image provides an accurate and convenient method for the study of temperature field in the actual welding process.更多还原