基于SVR的旋转电弧焊接形态预测模型(英文)

SVR for Prediction of Bead Geometry for Rotating Arc Welding

提出了SVR算法建立旋转电弧焊接形态的算法。旋转电弧焊接过程是一个多因素耦合的过程,有必要建立关于旋转电弧焊接的焊缝形态预测模型。选择旋转频率、旋转半径,焊枪高度和焊接电流的四个主要变量作为模型的输入,焊缝的宽度和深度作为输出。试验结果表明,基于SVR建立的焊缝形态预测模型在小样本训练数据的情况下具有较高的精度。

Using the SVR（ support vector regression） to build a model for prediction of bead geometry for rotating arc welding. The properties of rotating arc welding are affected by a large number of welding parameters, so modeling of weld bead shape is important for predicting the quality of weld. In this model, considering the effects of main variables as well as four factor interactions which are rotational frequency, rotational radius, height of torch and welding current are used as inputs, while the width and penetration of welding as output. The experiment resuits show that this model can achieve higher identification precision with a small size of training sample sets.