An inversion method for identifying the convection heat transfer coefficients of friction stir welding using a surrogate model

Shaofei Meng,Haitao Liu,Wei Yue,Juliang Xiao,Tian Huang,Yanbing Ni

DOI: https://doi.org/10.1007/s00170-023-11466-2

2023-05-03

Abstract:This paper focuses on identifying the convection heat transfer coefficients of the friction stir welding (FSW) temperature field from the temperature measuring data. Firstly, the numerical model of the FSW temperature field is constructed by utilizing the Lagrangian formulation. Then a novel inversion method combined with the surrogate model is proposed. The surrogate model based on proper orthogonal decomposition and radial basis function (RBF) is applied to approximate the input-output relationship of the numerical model. To improve the calculation efficiency and accuracy of the inversion method, an iterative updating method of the surrogate model combined with parallel computing technology is adopted. Subsequently, the effect of the type of RBF and the measurement position of the thermocouple on the calculation accuracy of the inversion analysis method are analyzed. The conclusions indicate that all the RBFs used in this paper are suitable for constructing the surrogate model of the temperature field. With the thermocouple measuring point close to the weld center, the maximum RMSE calculated by the numerical model decreases from 10.8 °C to 7.0 °C, but the maximum relative error of the peak temperature changes slightly, about 7.5%. The results illustrate that the identified convection heat transfer coefficients allow predictions on the temperature field of FSW which are consistent with the experimental data.

engineering, manufacturing,automation & control systems