基于熔池动态形状特征的熔透规律建模及控制

The welding penetration problem is a key problem affecting the welding quality and it is one of the factors restricting the development of welding automation. In view of the fact that a skilled welder can make accurate penetration judgment in the process of welding, the vision sensing method is used to capture the image information of the molten pool, and modeling and analysis of 3-D sensing of the molten pool shape characteristics will be carried out to determine the dynamic change trend of the molten pool shape characteristics. Then a nonlinear processing method will be used to determine the mapping relationship between the dynamic change trend and the quantitative index indicating the penetration status. Experiences of skilled welders are quantified to confirm the penetration criteria, which provide the basic data for the penetration control. The similarity coefficient model of the molten pool shape is to be established, and the relation between the similarity coefficient and the molten pool position is confirmed to provide feedback information for the molten pool positioning, then the consistency of the dynamic characteristic information of the molten pool shape is ensured. The impact factors of the main welding parameters on typical molten pool shape characteristic parameters are refined, and the corresponding quantitative relationship between the molten pool shape characteristic parameters and the welding parameters is determined. Combined with the penetration criteria, the key welding parameters are adjusted in real-time to ensure the welding penetration stability, and the root welding penetration problem can be solved with the welding quality improved.

焊缝熔透是影响焊接质量的关键问题，也是制约自动化焊接的发展要素之一。鉴于熟练焊工焊接时能够对熔透性作出准确判断，本项目拟采用视觉传感的方法，提取熔池图像信息，进行熔池形状特征的三维传感建模与解析，确定熔池形状特征动态变化趋势。在此基础上，用非线性处理方法确定熔池形状特征动态变化趋势与表征熔透状态的量化指标之间的映射关系。将熟练焊工经验量化，确定熔透判据，为熔透控制提供基础数据。建立熔池形状相似系数模型，并确定相似系数与熔池位置的对应关系，为熔池定位提供反馈信息，以确保熔池形状动态特征信息的一致性。提炼出主要焊接参数对典型熔池形状特征参数的影响因子，确定熔池形状特征参数与焊接参数之间的对应量化关系，结合熔透判据，对关键焊接参数实时调节，以保证焊接过程熔透稳定，解决根焊熔透难题，提高焊接质量。

熔透状态自动控制是实现打底焊接自动化的关键技术之一，由于焊接过程是多因素耦合的非线性随机波动过程，增加了熔透状态判定及自动控制的难度，目前相关研究仍处于理论探索阶段。通过双目视觉传感方法获取熔池特征信息并通过神经网络预测熔透状态与熟练焊工判断熔透状态方式较为一致，因此研究以双目视觉传感方法获取的熔池形状特征参数为输入量建立神经网络熔透预测模型，并设计出熔透状态自动控制系统，为打底焊接自动化的发展提供理论基础。.建立了双目视觉传感焊接熔池特征参数动态检测试验平台，提取GMAW熔池二维及三维形状特征参数以及背面熔宽信息，确定熔池形状特征参数随焊接参数的动态变化趋势。熔池背面熔宽是表征熔透状态最直接的特征信息，以熔池形状特征参数为输入、背面熔宽为输出建立熔透预测模型，并确定每个形状特征参数对于背面熔宽的权重系数。熔透状态预测模型代替焊工完成熔透状态判定，为熔透自动控制提供了可靠的传感反馈信息。建立熔池形状偏差系数模型，确定偏差系数随焊丝相对于坡口中心偏离量的变化趋势，为焊丝（熔池）对中调节提供有效参考信息。建立焊接参数与焊缝背面熔宽之间的数学模型，为熔透状态的控制调节提供基础数据。以背面熔宽的改变量作为输入，分别以焊接电流的调节量和焊接速度的调节量作为输出建立两个单输入单输出的自适应神经模糊系统，实现熔透控制焊接参数的自动调节。