文献类型：期刊文献

英文题名：Droplet transfer model for laser-enhanced GMAW

作者：Wang, XueWu[1,2];Huang, Yi[2];Zhang, YuMing[2]

机构：[1]E China Univ Sci & Technol, Coll Informat Sci & Engn, Shanghai 200237, Peoples R China;[2]Univ Kentucky, Dept Elect & Comp Engn, Lexington, KY 40506 USA

年份：2013

卷号：64

期号：1-4

起止页码：207

外文期刊名：INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY

收录：;EI(收录号：20130415930813);WOS:【SCI-EXPANDED(收录号:WOS:000313410500020)】；

基金：This work is funded by the National Science Foundation under grant CMMI-0825956 entitled "Control of Metal Transfer at Given Arc Variables".

语种：英文

外文关键词：Laser-enhanced GMAW; Metal transfer; Softsensing; BPNN; TSCPSO

摘要：Laser-enhanced gas metal arc welding (GMAW) is a recent modification of conventional GMAW. It applies a low power laser to the droplet to obtain an auxiliary detaching force to help the detachment and achieve controlled metal transfer. As a primary parameter that affects the process and weld quality, the size of the droplet needs to be monitored and controlled. However, its direct measurement requires a high-speed camera and is not preferred in a manufacturing site because of its high cost and complicated system structure. Soft-sensing method was thus proposed as an alternative to obtain it in real time. Laser power intensity, wire feed speed and welding voltage were identified as the major parameters that determine the droplet size and were thus selected as the auxiliary variables to estimate the primary parameters: size and transfer rate of the droplet. Least squares (LS) regression equation, back-propagation neural network (BPNN), particle swarm optimization-based back-propagation neural network, and three swarm cooperative particle swarm optimization-based back-propagation neural network (TSCPSO-BPNN) models were established from the collected data. Simulation results were analyzed and compared among these models. It was found that the selected auxiliary variables were closely related to the primary variables. Droplet size and transfer rate estimates made based on the TSCPSO-BPNN model are similar to those based on LS regression equation. After dimension reduction, the LS equation can be simpler than TSCPSO-BPNN model while the accuracy is sufficient and meets the requirement for future control. The estimation by LS method could thus be utilized in the control of the laser-enhanced GMAW.