Vibro-acoustic fault detection using a numerical time reversal method with reduced order models

In recent years, time reversal has been recognized as a potentially effective tool for structural health monitoring. By exploiting the property that the time reversed wave should converge to the original source location, a defect can be detected by sending the time reversed output data through a set of reference models with different fault locations. In this work simulated time reversed acoustic pressure and structural vibration signals are used, resulting from a structural excitation and a numerical model is employed to determine the position of a fault in the structure. The model has been built up with finite and infinite elements to account for the Sommerfeld radiation condition. To overcome the high computational power required to apply time reversal on all the reference models, model order reduction techniques are used and the time reversed simulations are performed on the reduced models. The chosen model order reduction procedure is a stability preserving second order Arnoldi Krylov subspace reduction. Recently it has been shown that this calculation method can retain the high-fidelity and stability of the finite element models in the required frequency range, but at a fraction of the computation cost of the full model. It is thus a viable tool for near real-time fault detection using numerical time reversal with the finite element method. Numerical results are compared for different fault locations and fault sizes and for different combinations of acoustic and structural sensor locations.

ISBN:9781510827936

Publication year:2016

Accessibility:Closed