Innovative data analysis techniques for structural health monitoring

Abstract

© 2017 International Society for Structural Health Monitoring of Intelligent Infrastrucure. All rights reserved. Many sources of uncertainties, which could be introduced into the structure during their construction and operational stages, make the reliable structural health monitoring (SHM) difficult. Environmental effect, finite element modelling error, methodology error and noise in the measurement data are the most significant types of uncertainties. The accuracy of structural condition assessment and the reliability of SHM results may be significantly affected. Besides, different data analysis techniques and damage indices may result in different structural damage identification results, even from the same measured data. The primary challenge of SHM is therefore development of methodologies that are sensitive to minor structural condition changes but insensitive to the changes by uncertain effects. This paper discusses and introduces a few signal processing techniques for SHM, including both the global vibration and local wave propagation based methods, to achieve a better accuracy and reliability of structural damage identification. The damage indices based on these new signal processing techniques are less susceptible to the uncertainties, but more sensitive to minor structural damage. These include: 1) Vibration Phase Space Topology. A damage index named change of phase space topology derived from the measured vibration responses is defined to detect the location of structural damage; 2) Phase Trajectory Change. The phase trajectories of multi-type vibration responses are obtained from a bridge under moving loads, and a damage index is defined as the separated distance between the trajectories of undamaged and damaged structures to detect the damage location; 3) Chaotic system based technique. A novel technique based on analyzing the responses of a nonlinear oscillator, i.e. Duffing-Holmes system, with the recorded ultrasonic guided wave signal as an added input is proposed for the detection of minor structural damage. Experimental verifications on various structures have been conducted to validate the accuracy and reliability of the proposed approaches to analyse the measurement data and detect the minor structural damage. The results demonstrate that these approaches are sensitive and accurate to detect the structural damage, but not sensitive to uncertainty effects.