Optimization of stacking sequence for quadrilateral laminated composite plates with curved edges based on Kriging

Weiping Wang,Qingshan Wang,Rui Zhong,Xianjie Shi,Liming Chen

DOI: https://doi.org/10.1016/j.camwa.2024.02.002

IF: 3.218

2024-02-16

Computers & Mathematics with Applications

Abstract:A Kriging-based optimization method is proposed and applied to the design optimization of stacking sequence for quadrilateral laminated composite plates with curved edges. The vibration model for evaluating the fundamental frequency of plates is constructed using the first-order shear deformation theory (FSDT). Then a Kriging surrogate model is used to approximate the vibration model to improve design optimization efficiency. A starting strategy based on the fitting quality coefficient is proposed to determine when to use the expected improvement (EI) infill criterion during the sequential sampling process. Three widely used correlation functions are dynamically selected during the Kriging modeling process to improve the generalization ability of model. In addition, a hybrid whale optimization algorithm (HWOA) is employed to optimize the Kriging hyperparameters, maximize the mean squared error (MMSE), the Expected Improvement (EI) function, and the Kriging predictor. Finally, a comparison is made with the classical DACE toolbox, the optimization procedure using the MMSE criterion and the standard one-shot Kriging-based optimization procedure. Results have demonstrated the effectiveness of the proposed method and clearly indicate that the required number of samples for optimization can be effectively reduced when starting to use the EI infill criterion at the fitting quality coefficient larger than 0.85.

mathematics, applied