As an effective welding method, rotating arc narrow gap MAG welding must ensure sidewall penetration, which is hard to realize when the groove is irregular in practical production. Therefore, it is significant to construct a weld form modeling for control weld penetration under the variable groove conditions. At first, the welding experiment system and the sensor system were set up. Arc sensing and visual sensing were applied to monitor the welding process, and effective algorithms were developed to extract the weld groove features. Three typical groove forms were considered and sufficient experiments were carried out to obtain enough experimental data. Furthermore, support vector machine is used to model the process. Welding current and the shortest distance between the arc center and the groove sidewall were used to predict the sidewall penetration. The verification results show that the model is effective under the condition that the height change of the groove is not more than 2 mm, the width change is not more than 2 mm, or the inclination angle of the bottom of the groove is not more than 10 degrees. The establishment of this model is very helpful to further control the formation of narrow gap welding.

作为一种有效的焊接方法，旋转电弧窄间隙MAG焊接必须确保侧壁穿透，而在实际生产中当沟槽不规则时很难实现。因此，构建用于控制可变坡口条件下的焊缝熔深的焊接形式建模非常重要。首先，建立了焊接实验系统和传感器系统。电弧感应和视觉感应被应用于监测焊接过程，并开发了有效的算法来提取焊接坡口特征。考虑了三种典型的凹槽形式，并进行了足够的实验以获得足够的实验数据。此外，使用支持向量机对过程进行建模。焊接电流和电弧中心与凹槽侧壁之间的最短距离用于预测侧壁的熔深。验证结果表明，该模型在凹槽高度变化不大于2mm，宽度变化不大于2mm，凹槽底部倾斜角不大于2mm的条件下有效。 10度 该模型的建立对进一步控制窄缝焊接的形成非常有帮助。