The Prediction Of Weld Bead Shape For High-Sthrength Steel By Laser-Arc Hybrid Welding

In the process of laser- arc hybrid welding, the weld profile is an important quality of weld quality evaluation. However there are too much controllable parameters in the process of the laser- arc hybrid welding,and any changes of the parameter would cause the change of the weld profile and then lead to the failure of welding.It is very worthy of study to find out the relationship between the parameters of Laser-Arc hybrid welding and the weld profile and to make prediction of the weld profile.First of all,The experiments is designed based on the response surface method, using obtained beam appearance data and filtering out significant factors by step wise regression method to establish multiple nonlinear mathematical regression model. By analysis of variance and regression analysis showed that R2 the regression model are 0.932, 0.915 and 0.910, P>F<0.001, respectively. The weld bead geometry was affected by the laser power, welding current, arc voltage and distance between two sources and these factor’s interaction. The main effect on penetration is the laser power, while the interactive effect is laser power and arc voltage; the main effect on the weld width is welding current and arc voltage, while the interactive effect is welding current and distance between two sources, arc voltage and distance between two sources, laser power and arc voltage; the main effect on the reinforcement is welding current, while the interactive effect is arc voltage and distance between two sources. The results show the simulating results are consistent with the measured results.According to the experiment and analysis of the impact of process parameters on the stability of the welding,we could find out that arc energy is the most important factor affecting the stability of the welding process,that is, the more stable the greater arc energy.At the same time the high-speed cameras was used to capture the arc shape,droplet transfer forms and the pool status to help study the relationship between the process parameters and the welding profile.And then,the weld profilometry data be collected by the polar coordinate system,and be fitted based on the profilometry data using MATLAB cubic spline.After analyzing the impact of various process parameters on the weld appearance by the mathematical model,the prediction model of the laser power, welding current, arc voltage,and the heat pitch to the profilometry be established using BP neural network.The results show that more than 90% relative error of prediction data and the actual data are within 20%.In order to improving the prediction accuracy and the generalization ability of BP neural network,the genetic algorithm to optimize was used to optimize the initial weights and thresholds of BP neural network.After a genetic algorithm optimization, the generalization and the prediction accuracy of neural network was improved. The results show that the relative error of all forecast data and actual data are less than 20%.