|
| Global sensitivity in nonlinear stochastic dynamic response analysis of structures |
| Revised:December 07, 2005 |
| View Full Text View/Add Comment Download reader |
| DOI：10.7511/jslx20082034 |
| KeyWord:structure,nonlinear,subset of the parameters,global stochastic sensitivity,probabimity density evolution method |
| CHEN Jian-bing LI Jie |
|
| Hits: 1577 |
| Download times: 9 |
| Abstract: |
| Measured by the relative contribution to the total variance of the target functions,a family of global stochastic sensitivity indices of random parameters or subset of the random parameters is defined.The Sobol' sensitive indices and their improvements,based on orthogonal decomposition of target function and according to partition of the total variance,are firstly revisited.As a family of global sensitivity indices,the idea of partition of the total variance is employed but implemented in a different way,yielding a new family of global sensitive indices.It is proved that this family of indices is essentially based on partially orthogonal decomposition of the target function.Combining with the probability density evolution method for nonlinear stochastic response analysis of structures,computational algorithm of the global stochastic sensitivity indices is outlined.The proposed iodices are of clear physical sense and conveniently computable.A hysteretic structure is taken as an example.The investigations indicate that the global stochastic sensitivity indices of different parameters are quite distinct to different target functions.Moreover,in some cases,in contrast to possible intuition tie randomness of parameters(or subset of parameters) may contribute negatively to the total variance of the target functions. |
|
|
|