Multi-objective Optimization of GMAW Short-circuit Transfer Process Based on Weighted Principal Component Analysis
The work aims to delve into the influence of gas metal arc welding(GMAW)process parameters on the forming quality of welded joints,so as to enhance the forming quality of Q235 steel during short-circuit transfer welding.To achieve this,orthogonal experiments and principal component analysis(PCA)were combined to examine the effects of short-circuit transfer welding process parameters on the aspect ratio,dilution rate,number of humps,and hardness gradient of the welded components.Subsequently,the process parameters were optimized.A multi-objective optimization approach was implemented for the weld bead with the maximum aspect ratio,minimum dilution rate,minimum number of humps,and minimum hardness gradient as optimization targets.This was achieved by calculating the weighted comprehensive evaluation values of the principal compo-nents.Notably,the response values derived from the first and second principal components accounted for 89.428％of the origi-nal data's information.Further,a variance analysis was conducted on the signal-to-noise ratio of the weighted principal compo-nent comprehensive evaluation values,leading to the determination of the optimal combination of process parameters:welding current of 120 A,welding voltage of 22 V,welding speed of 60 cm/min,and travel angle of 60°.These optimized welding condi-tions were then validated through experimental means,revealing a good consistency between the predictions and experimental results.The methodology employing weighted principal component analysis effectively optimizes the short-circuit transition process of GMAW while retaining a significant portion of the original data's characteristic information.Ultimately,the research findings contribute to the theoretical basis for optimizing the short-circuit transfer process in GMAW.
gas metal arc welding(GMAW)Taguchi methodshort-circuit transferprincipal component analysismulti-objective optimization
万方数据
维普
