Kriging-based uncertainty optimization of vibration characteristics for laminated elliptical shells considering material and load uncertainties

Abstract

In this paper, an uncertainty optimization method based on Kriging surrogate model is proposed to optimize the laying angles of laminated elliptical shells. First, a transient dynamic model of laminated elliptical shells is constructed to calculate the vibration energy. The accuracy of the transient dynamic model is validated by comparing the results from literature with the solutions of finite element model. Then, the Subtraction-Average-Based Optimizer is improved and used for the optimization of the hyper-parameters of the Kriging surrogate model. Furthermore, a sensitivity analysis is conducted using the constructed Kriging surrogate models to identify several uncertainty parameters that have a significant impact on the vibration energy response. Subsequently, two Kriging surrogate models with the identified significant uncertainty parameters and the design variables (laying angles) as input and the vibration energy as output are reconstructed for different structural boundaries, thicknesses, and shift distance of revolution axis, respectively. The applicability of these Kriging surrogate models for uncertainty analysis is verified by comparing with the Monte Carlo simulation results. Finally, the improved Subtraction-Average-Based Optimizer (ISABO) combined with the Kriging surrogate models is employed to optimize the laying angles of the laminated elliptical shells under material and load uncertainties. The results of the optimized peak values of vibration energy and intervals of peaks demonstrate that the uncertainty optimization method proposed in this paper is applicable and efficient.