Wind Turbine Blade Damage Detection and Localization via Spatiotemporal Kalman Filtering

Damage detection and localization remains among the most important and active research problems in vibration-based Structural Health Monitoring (SHM). Practical limitations call for a balance between the extraction of reliable structural response information and the density of the deployed sensor network. This study attempts to address this issue via incorporation of a spatiotemporal estimation framework, which is based on the effective combination of Kriging and Kalman filtering. While this strategy is quite popular in geostatistical, meteorological and environmental sciences, it has received considerably less interest in the field of structural dynamics. To this end, the proposed methodology assumes availability of data acquired from an observable and spatially continuous process. This allows the formulation of a spatiotemporal model aiming at describing the structure in its healthy state. Damage may then be detected and localized via the discrepancies between the spatiotemporal model estimates and the actual response data. The framework is implemented on simulated data from a finite element model of an experimental wind turbine blade, comprising E-glass fibres and a polyester resin as matrix material, subjected to dynamic loads. The proposed approach is validated on typical damage scenarios that pertain to certain stiffness reduction in damage prone areas of the blade. The obtained results indicate that spatiotemporal estimation may indeed serve as a significant tool towards damage localization, accounting for sparsity of the deployed sensor grids as well as potential sensor malfunctions of the network.