基于Bayesian理论的弯剪破坏钢筋混凝土柱变形能力概率模型

PROBABILISTIC DEFORMATION CAPACITY MODELS OF REINFORCED CONCRETE COLUMNS FAILED IN FLEXURAL-SHEAR BASED ON BAYESIAN THEORY

为了准确预测地震作用下弯剪破坏钢筋混凝土(RC)柱的变形能力,基于Bayesian理论提出了一种建立柱类构件变形能力概率模型的方法。首先对已有RC柱变形能力公式进行评价并选作先验模型,再借助美国PEER柱抗震性能试验数据库中发生弯剪破坏的20根矩形截面RC墩柱的拟静力试验数据作为客观认识,应用Bayesian参数估计方法综合这两类信息进行推断,修正先验模型的偏差从而建立反复荷载作用下弯剪破坏RC柱变形能力的概率模型;采用Bayesian方法剔除影响修正效果不显著的因素以获得模型简化,达到对先验模型的动态更新。最后,讨论了不同形式修正项对修正效果的影响。结果表明,Bayesian方法既继承了先验信息的完备性,又利用了试验数据的准确性,建议的概率模型较已有公式能够更准确地预测弯剪破坏RC柱的变形能力,从而为RC柱抗震设计或评估中考虑能力不确定性提供了便利。

In order to predict the deformation capacity of reinforced concrete columns(RC)failed in flexural-shear under seismic loading,a methodology to construct probabilistic deformation capacity models of columns is developed based on Bayesian theory.The prior models are proposed based on evaluating the biases of base deterministic models.And,the quasi-static test data of 20 rectangular RC columns failed in flexural-shear were collected from PEER-Structural Performance Database and used as objective knowledge.Two types of information,epistemic information and experimental information,are synthesized by Bayesian estimation method and inferred to update prior models.The probabilistic models for the deformation capacity of RC columns failed in flexural-shear under cyclic loading are constructed.Subsequently,the probabilistic models are simplified by using the stepwise procedure to delete unimportant terms and the dynamic updating of prior models is achieved.Then,the effect of different correction items is discussed.The results show that the approach inherits the completeness of prior models and the accuracy of experimental data information.Therefore,it could more accurately predict the deformation capacity of RC columns failed in flexural-shear,which is convenient to consider the capacity uncertainty in the aseismic design or in the evaluation of RC column.