基于等离子焊接的增材再制造焊道尺寸预测模型  

Prediction Model of Bead Geometry of Additive Remanufacturing Based on Plasma Arc welding

作　　者：王凯博[1] 吕耀辉[1] 刘玉欣[1] 徐滨士[1] 林建军[1] 孙哲[1] WANG Kaibo YU Yaohui LILT Yuxin XU Binshi LIN Jianjun SUN Zhe(National Defense Key Laboratory for Remanufacturing Technology, Academy of Armored Force Engineering, Beijing 100072, China)

机构地区：[1]装甲兵工程学院装备再制造技术国防科技重点试验室,北京100072

出　　处：《热加工工艺》2016年第19期211-214,共4页Hot Working Technology

基　　金：国防973及军队科研项目

摘　　要：建立一个基于遗传算法的神经网络模型对焊道尺寸进行预测。输入层单元为电流、送粉速率、焊接速度、离子气流量;输出层单元为焊道宽度、高度、熔深。采用正交试验方法得出的25组数据作为训练样本,控制变量法得出的16组数据作为预测样本。结果表明:模型预测精度高、效率快,训练误差范围在4.1%内,预测误差范围在6.5%内。并提出将模型应用到单道多层和多道多层增材再制造工艺中的方法。A neural network model based on genetic algorithm was established to predict bead geometry. The input unit consists of current, feeding rate, welding speed and plasma gas flow rate. The output unit consists of width, height and penetration of the bead. 25 sets of data obtained from the orthogonal test were used as training samples, and 16 sets of data obtained by the control variable method were used as the prediction sample. The results show that the prediction accuracy and efficiency of the model are high, the training error range is within 4.1%, and the prediction error range is within 6.5%. The method is proposed to apply the model to additive remanufacmring of multi-layer in single pass and multi-pass.

关 键 词：遗传算法 神经网络 增材再制造 焊道尺寸 预测模型 

分 类 号：TG456.2[金属学及工艺—焊接]