Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/7391
Title: Structural condition assessment with incomplete noisy acceleration measurements
Authors: Liu, Kun
Keywords: Structural analysis (Engineering)
Hong Kong Polytechnic University -- Dissertations
Issue Date: 2014
Publisher: The Hong Kong Polytechnic University
Abstract: Structural condition assessment is an essential phase of work in structural health monitoring. It is easy and practical to measure the acceleration responses of the structure, but the measured acceleration responses are usually incomplete in practice. In this thesis, structural condition assessment methods with incomplete measured responses are studied, in which force identification techniques and system identification techniques with incomplete acceleration responses are discussed with new methods proposed for an improvement. To reduce the effect of the discretization error of existing time domain force identification methods based on the state space method, an explicit form of Newmark-{461} algorithm is developed and applied to force identification. When the sampling frequency decreases or the time duration of sample increases, state space method may be inaccurate due to discretization error. The inverse analysis based on the state space method may also be inaccurate with this limitation. The conventional Newmark-{461} time-stepping integration algorithm is transformed into an explicit expression for the solution of the Ax=b equations in the force identification. With the advantage of computation stability of the Newmark-{461} algorithm, its explicit form gives more accurate identified force time histories compared with those from the conventional state space method. This proposed force identification method is extended to form a core component in a new substructural condition assessment method. The inverse analysis of a complex structure suffers great difficulty because of the large number of components and degrees-of-freedom (Dofs) in the finite element model. Different substructural methods have been developed in the last two decades to overcome this problem with a much reduced structural model. The substructures are inter-connected at the interfacing Dofs and knowledge of the interfacing forces would be essential for an accurate inverse analysis. Different discretization errors of state space method when applied to substructural analysis lead to inaccurate interface force identification. The proposed force identification method is used for substructural interface force identification and with a response sensitivity method considering interface force sensitivity. In this substructural analysis, the response sensitivity is affected by the measurement noise, and therefore two sensitivity enhancement methods are developed with the following description: Both sensitivity enhancement methods are based on subspace projection. In the decomposition of each measured response of the structure, the trend components from singular spectrum analysis (SSA) may contain most of the structural information and less noise information. This property is confirmed with numerical examples in this Thesis, and the trend components of the measured acceleration responses are chosen to construct the subspace for the projection of the sensitivity equation. Other than the measured response of the structure, the vectors of sensitivity as well as the computed response vectors are also projected into the corresponding decomposition subspace. The projected identification equations are then used to detect local damages in the structure.
For a structural system with known excitations, the computed response from the original model is noise free, and the principal components of these computed responses are selected to construct the subspace for the projection of the sensitivity equation. An improved model updating procedure is proposed with the enhanced sensitivity after the projection to estimate the stiffness reduction of the structure. This proposed method is found effective with high level of noise in the acceleration measurement. However, the analytical responses are used as the projection basis and it may not work when there is no knowledge on the excitation. These two kinds of projection method are studied with numerical simulations, and a laboratory test is performed to verify their effectiveness. A seven-storey planar steel frame is test with hammer impact in the laboratory, and the two methods described above are used to identify the local damages in the test model. Based on the proposed force identification method and response sensitivity enhancement methods, an improved structural condition assessment method is studied with incomplete acceleration measurements with or without noisy excitations. The numerical simulation study shows that the proposed structural condition assessment method will give satisfactory assessment result.
Description: xx, 259 leaves : illustrations ; 30 cm
PolyU Library Call No.: [THS] LG51 .H577P CEE 2014 LiuK
URI: http://hdl.handle.net/10397/7391
Rights: All rights reserved.
Appears in Collections:Thesis

Files in This Item:
File Description SizeFormat 
b2762898x_link.htmFor PolyU Users203 BHTMLView/Open
b2762898x_ir.pdfFor All Users (Non-printable)3.33 MBAdobe PDFView/Open
Show full item record

Page view(s)

155
Last Week
2
Last month
Checked on Jan 15, 2017

Download(s)

113
Checked on Jan 15, 2017

Google ScholarTM

Check



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.