Abstract:

Structural health monitoring (SHM) relies on models to exploit damage-sensitive features in order to identify and characterize damage.  Mechanistic models that are able to incorporate damage behavior are preferred, as they can represent damage type and severity, as well as characterize the danger posed to the structure by hypothesized damage scenarios.  Fusion of data-driven models, realized from vibrational response data, with useful mechanistic models must be achieved in order to  make use of sensor data in this way.  In addition, low-order models are preferred for SHM owing to the need for such systems to operate autonomously and in near real time.  In this study, a mechanistic model, based on spinning finite elements with damped gyroscopic and coupled flexural torsional effects, is fused with a data-driven model, based on output-only cyclo-stationary stochastic subspace identification, for spinning beam structures using an adaptive simulated annealing algorithm.  The ultimate goal of the project is to incorporate fluid-structure interaction into the mechanistic model for use in load characterization and SHM of wind turbine structures.  Related work integrating novel 2D strain-sensing skins for damage detection is also presented.