摘要
通过对直缝焊管高频感应焊接过程机理的深入研究,确定了影响焊接质量最大的开口角度、电流频率和线圈到焊接V点距离这3个工艺参数,同时提取表征焊接质量的焊缝处沿钢管壁厚的温度差和焊接热影响区最大等效残余应力这两个预测目标,然后利用遗传算法优化的BP神经网络建立了上述工艺参数和预测目标之间的非线性映射模型。测试结果表明,预测精度在±5%以内,BP神经网络的泛化能力良好,可以应用于直缝焊管高频感应焊接质量的实际预测。
Based on in-depth study on longitudinal welded pipe high frequency induction welding mechanism, it determined 3 welding process parameters affecting welding quality greatly, such as max. opening angle, current frequency and the distance from coil to welding V point. Meanwhile, it extracted two forecasts, including temperature difference along steel pipe wall thickness in weld, which characterize welding quality, and the maximum equivalent residual stress in HAZ. Then using genetic algorithm to optimize the BP neural network to establish the nonlinear mapping model between process parameters and forecast target. Test results showed that the prediction accuracy is within ＆#177;5%, the BP neural network generalization ability is good, which can be applied to the actual forecast for longitudinal welded pipe high-frequency induction welding quality.
出处
《焊管》
2014年第5期24-28,共5页
Welded Pipe and Tube
基金
河北省自然科学基金资助项目(E2009000395)
关键词
焊管
直缝焊管
高频感应
神经网络
质量预测
welded pipe
longitudinal welded pipe
high-frequency induction
neural network
quality forecast
