|
| Parameter identification methodology of restoring force model based on Bayesian theorem |
| Received:June 18, 2012 Revised:September 25, 2012 |
| View Full Text View/Add Comment Download reader |
| DOI：10.7511/jslx201305005 |
| KeyWord:Bayesian theorem restoring force model parameter identification multi-grid composite walls prediction model error |
| Author | Institution |
| 刘佩 |
北京交通大学 土木建筑工程学院, 北京 |
| 袁泉 |
北京交通大学 土木建筑工程学院, 北京 |
| 魏庆朝 |
北京交通大学 土木建筑工程学院, 北京 |
|
| Hits: 2485 |
| Download times: 1690 |
| Abstract: |
| Restoring force model parameter identification method based on Bayesian theorem is proposed in this paper,which takes account of the model prediction error.Using the tested hysteretic data,not only the most probable value but also the quantitative uncertainty of the model parameters can be obtained.Take the tested hysteresis curves of multi-grid composite walls under low cyclic loadings for instance,a restoring force model for multi-grid composite walls is proposed.Effects of stiffness degradation,strength degradation after ultimate load,and pinching are taken into account.Bayesian computational frame for parameter identification of restoring force model is proposed.Negative log-likelihood function of model parameter vector is derived,and then expressions of most probable value and covariance matrix of the parameter vector can be derived.Restoring force model parameters of precast multi-grid composite wall specimen and cast-in-place multi-grid composite wall specimen are identified.Hysteresis curves obtained through the most probable values,and backbone curves obtained through the most probable values and covariance matrixes are compared with the corresponding test data,which validate the proposed method and the identification results.The updated probabilistic density function of model parameters can be used for seismic risk assessment in future. |
|
|
|