基于显示连通贝叶斯网络的桥梁系统可靠性评估与更新

结题报告

项目介绍

AI项目解读

基本信息

批准号：
51708545
项目类别：
青年科学基金项目
资助金额：
21.0万
负责人：
王其昂
依托单位：
中国矿业大学
学科分类：
E0806.工程建造与服役
结题年份：
2020
批准年份：
2017
项目状态：
已结题
起止时间：
2018-01-01 至2020-12-31

项目参与者：
东兆星；刘书奎；邱若华；庄树森；岳子翔；
关键词：
系统可靠性易损构件识别贝叶斯网络链式化系统性能建模方法压缩算法

项目摘要

In order to solve the system reliability analysis problem for the bridge system with new information updating, the project explores explicit connectivity Bayesian network (ECBN) formulations for the system performance modeling. The goal is to define the probabilistic mapping relationship between system performance and its constituent component damage state, with the consideration of the component-level failure sequence correlation, multi-dimensional performance limit state (PLS) correlation and randomness. Considering multiple failure modes for the BN root node, system identification is carried out to obtain the structural modal and physical parameters based on the field measured data and improved Hilbert-Huang Transform (HHT) theory. The multi-dimensional PLS function is constructed to estimate the prior probability for the BN root nodes. The conditional probability evaluation method based on analytic hierarchy process (AHP) is proposed to describe dependencies between system components. BN chain-structure rule and compression algorithm are proposed to reduce conditional probability data and improve the inference efficiency. Finally, the system reliability can be calculated through BN forward inference based on the component reliability. The method is also able to update bridge system reliability when new information for is available. In addition, the most vulnerable system components are identified through BN backward inference. The results can provide scientific support to ensure safety construction and effective operation of the important major bridge structures.

针对有信息更新的桥梁系统可靠性评估问题，本项目探索改进的显示连通贝叶斯网络系统性能建模方法，综合考虑构件损伤次序相关性、多维性能极限状态相关及随机性，建立桥梁整体性能与构件损伤状态的映射关系。考虑网络构件根节点多种失效模式联合效应，基于实测数据与改进的Hilbert-Huang变换开展桥梁系统识别获取结构模态、物理参数，构造多维性能极限状态方程，评估网络根节点先验概率；探索基于层次分析法的中间节点条件概率评估模型，获取构件间条件概率依赖关系，改进既有贝叶斯网络参数估计算法。提出贝叶斯网络链式化法则及压缩算法，减少大型贝叶斯网络条件概率数据量，提高推理效率。通过贝叶斯网络正向推理实现从构件可靠性到结构系统可靠性的概率评估；当获取新监测数据时，更新桥梁整体性能状态；利用贝叶斯网络反向推理开展系统易损构件识别，为重大桥梁系统的安全建设与有效运营提供科学支撑。

结项摘要

本项目探索了基于显示连通贝叶斯网络系统性能建模方法，综合考虑构件损伤次序相关性、多维性能极限状态相关及随机性，建立桥梁整体性能与构件损伤状态的映射关系。考虑网络构件根节点多种失效模式联合效应，提出基于改进的Hilbert-Huang变换开展桥梁系统识别方法，获取结构模态、物理参数，构造多维性能极限状态方程，评估网络根节点先验概率；提出了基于层次分析法的中间节点条件概率评估模型，获取构件间条件概率依赖关系。通过贝叶斯网络链式化法则，减少大型贝叶斯网络条件概率数据量，提高推理效率。借助贝叶斯网络正向推理实现从构件可靠性到结构系统可靠性的概率评估；当获取新损伤信息时，更新整体性能状态；利用贝叶斯网络反向推理开展系统易损构件识别，对该类构件进行重点监测与加固。