Skip to Main content Skip to Navigation

Techniques d'anormalité appliquées à la surveillance de santé structurale

Abstract : The paradigm of structural health monitoring is based on the development of reliable and robust indicators able to detect, locate, quantify and predict damage. Studies related to damage detection in civil engineering structures have a noticeable interest for researchers in this area. Indeed, the detection of structural changes likely to become critical can avoid the occurrence of major dysfunctions associated with social, economic and environmental consequences.Recently, many researches have focused on dynamic assessment as part of structural diagnosis. Most of the studied techniques are based on time or frequency domain analyses to extract compressed information from modal characteristics or based on indicators built from these parameters. These indicators have shown their potentialities, but the problem of their sensitivity, the necessity of a reference state, and their reliability in terms of detection probability and false alarm, still remains. Moreover, the use of raw dynamic measurements (especially if several measurement channels are considered) leads to the storage of large datasets.In this context, it is important to use techniques capable of dealing not only with raw data but also modal parameters in a practical and relevant way. In order to give some insights to this problem, appropriate representations have been developed to improve both manipulation and storage of data. These representations are known as og symbolic data fg. They are used to characterize the variability and uncertainty that exists within each variable. The development of new methods capable of dealing with this type of data is the goal of Symbolic Data Analysis (SDA).This thesis has two main objectives: the first one is to use different methods coupled with the SDA to detect structural damage. The idea is to employ clustering procedures (e.g., hierarchy-divisive, hierarchy-agglomerative and dynamic clouds) and supervised classification methods (e.g., Bayesien decision trees, neural networks and support vector machines) to discriminate different structural states. In this thesis, SDA is applied to dynamic measurements obtained on site (accelerations) and to the identified modal parameters. The second goal is to study the impact of environmental effects, particularly those related to thermal variation over modal parameters. To this end, a couple of regression techniques are proposed.In order to attest the efficiency of the proposed approaches, several sensibility studies considering numerical applications and experimental investigations are carried out. It is shown that SDA coupled with classification methods is able to distinguish structural conditions with adequate rates. Furthermore, it is stressed the importance of using techniques capable of correcting modal parameters from thermal effects in order to build efficient procedures for damage detection
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Thursday, March 31, 2011 - 4:56:29 PM
Last modification on : Friday, June 16, 2017 - 10:30:26 AM
Long-term archiving on: : Friday, July 1, 2011 - 2:31:31 AM


Version validated by the jury (STAR)


  • HAL Id : tel-00581772, version 1



Alexandre Cury. Techniques d'anormalité appliquées à la surveillance de santé structurale. Autre. Université Paris-Est, 2010. Français. ⟨NNT : 2010PEST1055⟩. ⟨tel-00581772⟩



Record views


Files downloads