TY - CONF

T1 - Bootstrap tests for model selection in robust vibration analysis of oscillating structures

AU - Kargoll, Boris

AU - Omidalizarandi, Mohammad

AU - Paffenholz, Jens-André

AU - Neumann, Ingo

AU - Kermarrec, Gael

AU - Alkhatib, Hamza

N1 - Funding information: The research was funded partly by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 386369985. In addition, the research was partly funded and carried out within the scope of the collaborative project "Spatio-Temporal Monitoring of Bridge Structures Using Low Cost Sensors" with ALLSAT GmbH, which was supported by the German Federal Ministry for Economic Affairs and Energy (BMWi) and the Central Innovation Programme for SMEs (Grant ZIM Kooperationsprojekt, ZF4081803DB6). The authors acknowledge the Institute of Concrete Construction (Leibniz University Hannover) for providing the shaker table and the reference accelerometer used within the experiment.

PY - 2019/5

Y1 - 2019/5

N2 - In this contribution, a procedure for deciding,whether the oscillation of a surveyed structure is damped or not, is proposed. For this purpose, two bootstrap tests under fairly general assumptions regarding auto-correlation and outlier-affliction of the random deviations (“measurement errors”) are suggested. These tests are derived from an observation model consisting of (1) a parametric oscillation model based on trigonometric functions, (2) a parametric auto-correlation model in the form of an autoregressive process, and (3) a parametric stochastic model in terms of the heavy-tailed family of scaled t-distributions. These three levels, which generalize current observation models for oscillating structures, are jointly expressed as a likelihood function and jointly adjusted by means of a generalized expectation maximization algorithm. Closed-loop Monte Carlo simulations are performed to validate the bootstrap tests. Visual inspection of models fitted by standard least-squares techniques are shown to be insufficient to detect a small significant damped oscillation. Furthermore, the tests are applied to a controlled experiment in a laboratory environment. The oscillation was generated by means of a portable shaker vibration calibrator and measured by a reference accelerometer and a low-cost accelerometer.

AB - In this contribution, a procedure for deciding,whether the oscillation of a surveyed structure is damped or not, is proposed. For this purpose, two bootstrap tests under fairly general assumptions regarding auto-correlation and outlier-affliction of the random deviations (“measurement errors”) are suggested. These tests are derived from an observation model consisting of (1) a parametric oscillation model based on trigonometric functions, (2) a parametric auto-correlation model in the form of an autoregressive process, and (3) a parametric stochastic model in terms of the heavy-tailed family of scaled t-distributions. These three levels, which generalize current observation models for oscillating structures, are jointly expressed as a likelihood function and jointly adjusted by means of a generalized expectation maximization algorithm. Closed-loop Monte Carlo simulations are performed to validate the bootstrap tests. Visual inspection of models fitted by standard least-squares techniques are shown to be insufficient to detect a small significant damped oscillation. Furthermore, the tests are applied to a controlled experiment in a laboratory environment. The oscillation was generated by means of a portable shaker vibration calibrator and measured by a reference accelerometer and a low-cost accelerometer.

KW - structural health monitoring

KW - low-cost accelerometer

KW - vibration analysis

KW - damped harmonic oscillation

KW - robust parameter estimation

KW - model selection

KW - bootstrap test

M3 - Paper

T2 - 4th Joint International Symposium on Deformation Monitoring

Y2 - 15 May 2019 through 17 May 2019

ER -