Structural health monitoring of VBI systems via signal processing of vibration data

This thesis presents a collection of innovative signal processing strategies to detect both location and severity of damage in bridge structures using moving mass experiments. The literature reviewed shows that parametric damage detection strategies require different types of information from the both intact and damaged structures, such as (1) a finite element model (FEM) of the intact structure, (2) data collected from the intact structure, or (3) data collected only from the damaged structure. Methods are proposed that exploit one or more of the aforementioned information types. These methods are tested using numerical Vehicle Bridge Interaction (VBI) models that consider the effect of road roughness, measurement noise and boundary conditions through Monte Carlo simulation. Analytical formulas are derived to calculate damage severity using data obtained from VBI models. The proposed methods are proven to be robust to as much as 100% noise in measurements and their superiority over other state-of-the-art methods from the literature is demonstrated. In particular it is noted that in most methods reported in the literature a constant moving load velocity is assumed, while the present work demonstrates the ability to locate and quantify the damage with a moving load of variable velocity. Finally, state-of-the-art real-time damage detection may use advanced sensing technologies. It is shown how a new piezo-floating-gate (PFG) sensor may be used in a damage detection strategy that outperforms other techniques proposed in the literature