Uncertainty Research Of Injection Mechanism Under Bayes-Kriging Model

The squeeze casting technology is a kind of new forming process which possesses high-quality forming,wide scope of application and high productive efficiency.The squeeze casting equipment is the core equipment platform for implementing this technology,its performance directly decides the technical level of squeeze casting technology.This article regards the injection mechanism of squeeze casting equipment as the main research object,analyzes the technological parameters,material properties,work condition and other uncertain influences which exist in the forming process of squeeze casting,optimizes the numerical simulation method and improves the reliability of forming equipment for purpose to develop the analysis of injection mechanism friction model finite elements,the uncertain prediction of Kriging surrogate model,the reliability of Bayesian method and the practical application of Bayes-Kriging method and other relevant researches.A dynamic friction method is presented based on the traditional model of thermo-mechanical coupled analysis.That is,the friction characteristics research by combining the lubrication model and friction test,and the update the iterative boundary conditions based on MSC.Marc user subroutine.The established friction model can be used to perform the coupling analysis of thermal-mechanical and friction.The reliability and accuracy of the dynamic friction model are proved by compared with the laboratory testing on 100 t press machine.The basic idea of meta-modeling and Bayesian method are introduced to the field of casting molding.A spatiotemporal multiple-inputs-single-output(MISO)predictive model is established,according to the temporal and spatial distribution characteristics of temperature in process,and quantitatively verify by Bayesian statistical inference method.Then analyzes the effect of threshold?on the accuracy of the model.The prediction results indicate that the possibility of uncertainty is high in the initial and gradually lower with the stable.Threshold?change has a significant influence on the normal data,both Bayes factor and confidence can accelerated growth in a smaller?range.Thus we can further control the accuracy of the model by controlling the size of threshold?.A generalized procedure of Bayes-Kriging model is presented to implement the proposed probabilistic methodology for uncertainty research of 2500 kN squeeze casting.The results show a great deal of uncertainty in the initial stage of forming cycle and the moment between each injection cycle.Hypothesis testing demonstrates that,the proposed model with high accuracy.Temperature prediction of new node by using Bayes-Kriging method,the maximum relative error is small,and the degree of discretization error is low.The results indicates that the Bayes-Kriging model has good accuracy and has certain feasibility in instead of inefficient finite element simulation method.