TWO-DIMENSIONAL ACOUSTIC EMISSION SOURCE LOCALIZATION ON A LAMINATED VENEER LUMBER PLATE BY GAUSSIAN PROCESS REGRESSION
Abstract
Acoustic emission (AE) source localization is an important part of monitoring the health of infrastructures. Though straightforward for isotropic materials, where analytical solutions exist, locating sources for anisotropic materials is complicated due to the angle-dependent wave velocity. The problem is even more intractable if heterogeneity involves. This is the case for laminated veneer lumber (LVL), an engineered wood material composed of multiple thin layers of wood, where defects such as knots, voids and discontinuity distributed randomly within the layers, greatly undermining the effectivity of common localization methods for man-made composites. To avoid the problem of heterogeneity, this work employed Gaussian process regression (GPR) to address the 2D AE source localization problem in an LVL plate. With four AE sensors attached at the corner of ROI in the plate, multiple pencil lead break tests were conducted to collect the difference of time of arrival (dTOA) between different sensors. The vector of dTOAs serves used as the input of the Gaussian process regression while the output is the source location. The marginal likelihood was maximized to achieve the optimal model parameters. The input vectors of a different combination of dTOA components were fed into the GPR model, both the predictions on grid points and off-grid points were analyzed. The high accuracy of the mean predictions over all possible combinations of dTOAs indicates such a method can well cope with 2D source localization problems in LVL structures.
Reference
XIANGDONG HE; XUAN ZHU. (2023). TWO-DIMENSIONAL ACOUSTIC EMISSION SOURCE LOCALIZATION ON A LAMINATED VENEER LUMBER PLATE BY GAUSSIAN PROCESS REGRESSION. Proceedings of the 14th International Workshop on Structural Health Monitoring. https://doi.org/10.12783/shm2023/36912
Explore the literature
Follow references, discover citing works and find related studies.