A particle filter method for damage location in plate-like structures by using Lamb waves

This paper proposes a probabilistic approach for localizing damage in plate-like structures by using particle filter (PF) method with considerations of the uncertainties from modeling and measurement. A piezoelectric sensor network is deployed on the plate-like structures to excite and receive diagnostic Lamb waves before and after damage. The damage-scattered waves that contain characteristic information about the damage are then obtained and used for damage localization. A complex Morlet wavelet transform is employed to highlight the time-of-arrival of the scattered waves in each actuator–sensor path to help measuring the time-of-flight values. After that, damage localization is formulated as a Bayesian state estimation problem with the damage location and the wave velocity as the unknown state parameters. A PF procedure is developed to iteratively estimate the unknown parameters simultaneously. Numerical study for an aluminum plate and experimental study for a stiffened aluminum panel are conducted to demonstrate the effectiveness and applicability of the proposed PF-based localization method.