摘要
为了研究冷金属过渡(CMT)脉冲焊接过程中工艺参数对熔透、焊缝熔宽和焊缝余高的影响,提出了一种用于预测焊缝几何形状(焊缝高度和宽度)和熔透状态(深度和面积)的反向传播神经网络模型。该模型以峰值焊接电流,焊接速度和热量输入作为输入参数,且以焊缝高度和宽度、熔深和稀释面积作为输出参数,并给出了设计框架。结果表明,神经网络与训练数据有很好的一致性,可以有效地用于焊缝和熔透几何参数的预估。预测值与实验值之间的误差百分比较小,验证了提出模型的有效性。
The effect of process parameters on penetration,welding width and welding residual height during the pulse welding of cold metal transition（CMT） were studied by a back propagation neural network model,which was proposed to predict the geometry of the weld and the penetration state.The peak welding current model was carried out,the welding speed and the heat input as input parameters,and the height and width of the weld,the depth of the weld and the area of the dilution were used as the output parameters,and the design frame was given.The results show that the neural network has good consistency with the training data and can be effectively used for the estimation of weld and penetration geometry parameters.The percentage of error between the predicted and experimental values is small,which verifies the validity of the proposed model.
作者
郭艳平
陈剑虹
侯凤贞
GUO Yanping;CHEN Jianhong;HOU Fengzhen(Jincheng College,Nanjing University of Aeronautics and Astronautics,Nanjing 211156,China;State Key Laboratory of Gansu Advance Non-ferrous Metal Materials,Lanzhou University of Technology,Lanzhou 730050,China;School of Science,China Pharmaceutical University,Nanjing 210009,China)
出处
《铸造技术》
CAS
2018年第11期2575-2579,共5页
Foundry Technology
关键词
人工神经网络
CMT
焊接参数
焊缝几何形状
焊透深度
回归模型
artificial neural network
CMT
welding parameters
weld bead geometry
penetration depth
regression model
