考虑替代模型不确定性的结构动力特性全局敏感性分析.

Uncertainty of structural parameters will unavoidably lead to uncertainty of structural natural frequencies. The global sensitvity analysis ( GSA) is an effective approach to quantify the contributions of individual parameters to the induced uncertainty of dynamic characteristics. However, the GSA has the issue of high computational cost that needs to be addressed. The fast-running Gaussian process model (GPM) was used as a surrogate for the costly computer models, to reduce the computational burden of the GSA. Moreover, the influence of the metamodel uncertainty associated with the GPM was taken into account. The effectiveness of the presented GPM-based method for the GSA was verified with a test function. Finally the GPM-based approach was applied to the GSA of structural dynamic characteristics of the Anqing Yangtze River Railway Bridge. [ABSTRACT FROM AUTHOR]