基于离散动态贝叶斯网络的桥梁状态评估方法

Method of Bridge Condition Assessment Based on Discrete Dynamic Bayesian Networks

针对现有桥梁结构状态评估方法不能通过检测信息进行更新的缺陷,提出基于离散动态贝叶斯网络的桥梁状态评估方法。以混凝土斜拉桥为研究目标,利用层次分析法建立桥梁状态评估体系框架,将评估体系中的各个构件元素设定为离散动态贝叶斯网络中的节点,并根据结构状态的划分等级赋予节点离散值域,按时间序列建立适用于桥梁结构状态评估的动态贝叶斯网络模型。对利用层次分析法产生的离散状态样本进行学习,得到模型参数。采用该方法对某座混凝土斜拉桥进行状态评估,验证了其有效性,并说明动态贝叶斯网络能利用已有的检测信息对桥梁各个构件的状态评估结果进行更新和修正。

Owing to the defects that the existing methods of bridge condition assessment can not update their models,using the detection information,a method of the bridge condition assessment based on the discrete dynamic Bayesian networks was proposed.The concrete cable-stayed bridges were taken as the research objects and the analytic hierarchy process method was used to establish the framework of the bridge condition assessment system.The component elements in the assessment system were set as the nodes in the dynamic Bayesian networks and in accordance with the classification of the structural conditions,the discrete value ranges were given to the nodes and the dynamic Bayesian networks model that was applicable to the bridge condition assessment was established in terms of the time series.Meanwhile,the discrete state samples generating from the analytic hierarchy process method were learnt and the parameters of the model were determined.The proposed method was used for the condition assessment of a concrete cable-stayed bridge,the effectiveness of the method was verified and the facts that the dynamic Bayesian networks could update and modify the condition assessment results of the different bridge components,using the existing detection information,were also illustrated.