【摘要】 不论熔透控制还是焊缝跟踪 ,都需要实时或离线提取焊接熔池正面或反面的几何参数 .在小电流脉冲 TIG焊接试验条件下 ,为了获得训练焊缝熔透神经网络模型的大量数据 ,提出了一套方法 ,包括用于实时提取焊接熔池正面几何参数的算法和离线提取焊缝背面熔宽的摄影测量法 ,并采用 C+ +语言开发了相应的软件 .试验证明 ,该方法是可行的更多还原

【Abstract】 It is necessary to extract molten pool geometrical parameters in real time or to extract back bead width off-line either in seam tracking or in penetration control. In order to obtain the mass data to train neural network model of seam penetration under the condition of low current pulsed TIG(tungsten inert gas) welding, a set of methodology including the algorithm for extracting the welding molten pool front-side geometrical parameters and the method for measuring the back bead width was proposed. The corresponding software was developed using C++ language. The experimental results show that the methodologies are feasible.更多还原