|
| Optimal probabilistic collocation points for stochastic response surface method |
| Received:April 22, 2011 Revised:November 12, 2011 |
| View Full Text View/Add Comment Download reader |
| DOI：10.7511/jslx20123009 |
| KeyWord:stochastic response surface method probabilistic collocation regression method linearly independent reliability |
| Author | Institution |
| 蒋水华 |
武汉大学 水资源与水电工程科学国家重点实验室; 水工岩石力学教育部重点实验室, 武汉 |
| 李典庆 |
武汉大学 水资源与水电工程科学国家重点实验室; 水工岩石力学教育部重点实验室, 武汉 |
| 周创兵 |
武汉大学 水资源与水电工程科学国家重点实验室; 水工岩石力学教育部重点实验室, 武汉 |
|
| Hits: 2043 |
| Download times: 1665 |
| Abstract: |
| The accuracy and efficiency of the SRSM based on the linearly independent probabilistic collocation points are studied systematically.The flow chart of the procedure based on principle of linearly independent probabilistic collocation points is proposed.Both the linearly independent probabilistic collocation points and the probabilistic collocation points using the regression method are compared.Two examples are presented to demonstrate that the selected number of collocation points should ensure the Hermite coefficient matrix having a full rank for SRSM based on the probabilistic collocation points using the regression method.In this situation,the accuracy of SRSM is sufficient enough.Furthermore,its efficiency is higher than the traditional Monte Carlo simulations.The SRSM based on the linearly independent probabilistic collocation points can result in the same accuracy as the traditional Monte Carlo simulations,and its efficiency is significantly higher than the SRSM with the probabilistic collocation points using the regression method,which may be taken as an alternative method for structural reliability analysis,particularly for the complex structural reliability problems with implicit performance function.These results can provide a basis for determining the optimal number of probabilistic collation points for the application of SRSM to structural reliability analysis. |
|
|
|