Research On Multi-Objective Optimization Of Laser Beam Oscillating Welding Process Parameters

Laser beam oscillating welding can stabilize the welding process and improve welding performances.But for the laser beam oscillating welding process,the performance of the weld is usually affected by the laser oscillating mode,laser power,welding speed and other process parameters.However,there is a highly complex relationship between welding process parameters and weld performances,so it is difficult to optimize the weld performances by directly adjusting the welding process parameters.Therefore,it is of great research value to explore theories and methods related to multi-objective optimization of process parameters based on experiments.This paper focuses on the laser beam oscillating welding of 6 mm thick 5083 aluminum alloy.First,the single-factor experiments of laser beam oscillating welding are carried out to study the effects of single welding process parameter(laser power,welding speed,focal position,oscillating amplitude,and oscillating frequency)on the welding performances(the depth of penetration,aka DP,the depth-to-width ratio of welding bead,aka DW,and the porosity of welding bead,aka PW).Then,the five-factor orthogonal experiments of laser beam oscillating welding are carried out to study the importance of welding process parameters on the welding performances through range analysis.The regression analyses of the functional relationship between the weld performance parameters and the process parameters are carried out by using the Kriging interpolation model,the error back-propagation neural network model,the radial basis function neural network model,and the generalized regression neural network model based on the experimental datas of laser beam oscillating welding.Then,the errors of every different models on the test set are calculated.Compared with the best performance of the independent models,the hybrid model improves the prediction performance of DP,DW,and PW by 31.72%,37.93%,and 27.9%,respectively.The multi-objective optimization model is established with the objective functions of DP,DW,and PW.On the basis of the hybrid model,the unified objective method and the NSGAII multi-objective evolutionary algorithm are used to optimize the multi-objective functions respectively.The results show that the results obtained by the NSAG-II algorithm are better than the unified objective method.