TIG弧钎堆焊熔池视觉检测与质量控制研究

【作者】 郭国林；

【导师】 王克鸿；

【作者基本信息】 南京理工大学 ， 材料加工工程， 2004， 硕士

【摘要】 在分析TIG弧钎堆焊电弧光谱的基础上，设计了不同组合的复合滤光系统，通过所取熔池图像对比，确定了最佳的取像窗口。采用不同的工艺规范进行了大量的实验，对铜带在不同熔化状态下的熔池图像进行了分析研究，得出了铜带全部熔化、部分熔化和过熔化状态下的熔池视觉图象特征，为钎堆焊熔深控制提供了技术基础。针对TIG弧钎堆焊铜熔池图像的特点，设计了相应的图像处理算法，提取出了熔池的几何形状参数，处理一幅图像整个过程不超过120ms。采用BP神经网络技术建立了焊接工艺参数、熔池正面几何形状参数与熔深之间的关系模型，为熔深模糊控制奠定了基础。设计出了以熔深的偏差和偏差变化率为输入、焊接电流的变化为输出的模糊控制器，对控制器进行了仿真和实际控制实验，结果表明，该控制器具有良好的控制效果，获得了满意的焊接质量。更多还原

【Abstract】 Based on the analysis of tungsten-inert-gas (TIG) welding arc spectrum, different composite light-filter systems is designed. After comparing the grabbed weld pool images, the optimal window for capturing the images is determined. By adopting different process criterion, carrying out great number of experiments and studying the weld pool images when copper-band is in different molten statuses to obtain the characteristics of weld pool visual images when copper-band is completely molten, partly molten or excessively molten, so provide technology foundation for the penetration control of braze-surface welding. According to the characteristics of copper weld pool images, a suitable image processing algorithm is developed and the geometrical parameters of weld pool are acquired. The whole processing time is less than 120ms. By means of BP neural network technology, a model describing the relationships between weld process parameters, the topside geometrical parameters of weld pool and weld penetration is established. The model lays a solid foundation for fuzzy-control of weld penetration. A fuzzy intelligent controller is designed with weld penetration warp and its variation as inputs, the variation of welding current as output. The simulation and control experiment show that this controller has good control results and obtains satisfying welding quality.更多还原