Compressive sensing with an adaptive wavelet basis for structural system response and reliability analysis under missing data. (1st April 2017)

• Record Type: Journal Article
• Authors: Comerford, L.; Jensen, H.A.; Mayorga, F.; Beer, M.; Kougioumtzoglou, I.A.
• Is Part Of: Computers & structures. Volume 182(2017)
• Page Start: 26
• Page End: 40
• Publication Date: 2017-04-01
• Subjects: Compressive sensing -- Harmonic wavelets -- Power spectrum -- Missing data -- Advanced simulation techniques -- Reliability analysis; Structural engineering -- Data processing -- Periodicals; Electronic data processing -- Structures, Theory of -- Periodicals; 624.171
• DOI: 10.1016/j.compstruc.2016.11.012
• Languages: English
• ISSNs: 0045-7949

Abstract

Highlights: Earthquake excitation power spectra are estimated from incomplete process records. Process records with missing data are reconstructed using compressive sensing. Iteratively re-weighted harmonic wavelet bases are utilized for EPS estimation. A reinforced concrete building FEM is subjected to earthquake ground accelerations. System response and reliability statistics are estimated with up to 40% missing data. Abstract: The challenge of determining response and reliability statistics of large-scale structural systems under earthquake induced stochastic excitations is considered where the source load data records are incomplete. To this aim, a compressive sensing based framework in conjunction with an adaptive wavelet basis is presented for reconstructing the samples with missing data and estimating the underlying process EPS. In this regard, novel insights are provided whereas certain conceptual, numerical, and practical implementation aspects of the technique are presented in detail. A numerical example pertaining to the stochastic response and reliability analysis of an eight floor reinforced concrete building structural system demonstrates the effectiveness of the proposed methodology.