Resumen
Due to manufacturing errors, inaccurate measurement and working conditions changes, there are many uncertainties in laminated composite cylindrical shells, which causes the variation of vibration characteristics, and has an important influence on the overall performance. Therefore, an uncertainty representation methodology of natural frequency for laminated composite cylindrical shells is proposed, which considers probabilistic and interval variables simultaneously. The input interval variables are converted into a probabilistic density function or cumulative distribution function based on a four statistical moments method, and a unified probabilistic uncertainty analysis method is proposed to calculate the uncertainty of natural frequency. An adaptive Kriging surrogate model considering probabilistic uncertainty variables is established to accurately represent the natural frequency of laminated composite cylindrical shells. Finally, the dimensionless natural frequency of three-layer, five-layer and seven-layer laminated composite cylindrical shells with uncertainty input parameters is accurately represented. Compared with the Monte Carlo Simulation results, the mean error and standard deviation error are reduced to less than 0.07% and 4.7%, respectively, and the execution number of calculation function is significantly decreased, which fully proves the effectiveness of the proposed method.