Robust Optimization of Dynamic Fit Parameters for Injection Mechanism in Squeeze Casting Based on Six-sigma Frame and Bayesian Kriging Metamodel

Authors：You, Dongdong^*; Luo, Xiao; Zhu, Yanghui; Deng, Jianxin

Source：International Journal of Metalcasting, 2023, 17(1): 124-142.

DOI：10.1007/s40962-022-00756-2

• robust optimization
• kriging metamodel
• uncertainty quantification
• injection mechanism
• bayesian inference

Robust optimization of injection mechanism can improve the stability of injection process control. A novel multiobjective six-sigma (6 sigma) robust optimization method that considers key indices based on Bayesian Kriging metamodel was proposed in this study. The deformation and temperature of the injection mechanism were studied, and the stability was characterized using the clearance rate. By combining a Kriging metamodel and Bayesian hypothesis testing, an approximate system exhibiting high reliability was established. Based on the approximate system, a robust optimization design of the injection parameters was proposed under the 6 sigma framework. The optimization results of multiobjective particle swarm optimization and NSGA-II with different population sizes and generations numbers are compared. Results showed that NSGA-II (population size = 32; generations number = 60) gave the better performance. Robust optimization on this basis reduced the average clearance rate and standard deviation of the injection mechanism by 11.38% and 30.8%, respectively.

• Institution
  广西大学