大功率光纤激光焊焊缝识别回归算法研究

Study of regression algorithm for detecting weld during high-power fiber laser welding

针对工业实际焊接加工过程的焊缝偏差,利用IPG YLR-10000大功率光纤激光焊接设备,通过高速视觉传感器获取动态熔池红外图像序列图,观察激光束偏离焊缝时熔池红外辐射的特征变化,采用焊接熔池前峰点位置作为焊点偏移特征参数测量值,试验证明该参数能体现激光束偏离焊缝的实际情况。最后,利用回归分析方法建立焊点实际偏差与焊点位置测量偏差的红外动态视觉模型,通过分析该模型的相关性参数、对F检验法中F参数的结果进行验证,确定了该二元回归建模的有效性及可行性。

To detect the weld seam deviation during the industrial actual welding process, an IPG YLR-10000 high power fiber laser welding equipment was used to obtain a dynamic pool infrared image sequence diagram through a high-speed visual sensor. After observing the weld pool infrared radiation characteristics changing when the laser beam deviated from the weld, the prior point of the weld seam pool which could be tested to reflect the actual position of the deviation of the weld, could be chosen as the feature measurement parameter. Finally, a model of the relationship of the actual weld deviation and the tested weld deviation had been set up by using the regression analysis method. Its validity and feasibility had been tested by analyzing the model correlation parameters.