Structural reliability updating and damage assessment with Bayesian networks

Haifeng Yang, Ziyan Wu, Hongbin Sun, Peng Sun

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The traditional reliability analysis method cannot handle discrete variables and cannot give response timely when there are a new information (usually measurements) about the structure. In this paper, a computational framework for structural reliability updating and damage assessment is proposed. System modeling of traditional Bayesian network is developed with continuous variables discretization and elimination. Firstly the reliability Bayesian network (RBN) is established according to the structure type, and to discretize the key component information variables. Then the redundant continuous nodes are eliminated according to Shachter's node elimination rule. Final the variable elimination method is used to carry out the exact inference and calculate the posterior probability distribution. When evidence (measurement) is available, at forward inference, the proposed method can facilitates structural reliability updating, and at backward inference, the proposed method can carry out damage assessment for the selected key component. Taking a rigid frame as the research object. By comparison with Monte Carlo method, the reliability deviation is less than 5%, it shows validity and accuracy of the proposed method.

Original languageEnglish
Title of host publicationStructural Health Monitoring 2017
Subtitle of host publicationReal-Time Material State Awareness and Data-Driven Safety Assurance - Proceedings of the 11th International Workshop on Structural Health Monitoring, IWSHM 2017
EditorsFu-Kuo Chang, Fotis Kopsaftopoulos
PublisherDEStech Publications
Pages1154-1161
Number of pages8
ISBN (Electronic)9781605953304
DOIs
StatePublished - 2017
Event11th International Workshop on Structural Health Monitoring 2017: Real-Time Material State Awareness and Data-Driven Safety Assurance, IWSHM 2017 - Stanford, United States
Duration: Sep 12 2017Sep 14 2017

Publication series

NameStructural Health Monitoring 2017: Real-Time Material State Awareness and Data-Driven Safety Assurance - Proceedings of the 11th International Workshop on Structural Health Monitoring, IWSHM 2017
Volume1

Conference

Conference11th International Workshop on Structural Health Monitoring 2017: Real-Time Material State Awareness and Data-Driven Safety Assurance, IWSHM 2017
Country/TerritoryUnited States
CityStanford
Period09/12/1709/14/17

Funding

The support from National Science Foundation of China (No. 51278420) is greatly appreciated. This research was also supported by Natural Science Foundation of Shaanxi Province of China (No. 2017JM5021).

Fingerprint

Dive into the research topics of 'Structural reliability updating and damage assessment with Bayesian networks'. Together they form a unique fingerprint.

Cite this