UC San Diego

The incomplete measurement problem poses a significant obstacle for both model correlation and structural health monitoring (SHM). In practice, information about the health of a structure must be ascertained using measurement data from only a limited number of sensors. Several approaches to this problem have been proposed. One approach involves a reduced order model, such as that obtained using methods such as the Guyan reduction, the Improved Reduced System Model, or the System Equivalent Reduction Process. A second approach, which this work considers, involves an expansion of the test data to match a higher-fidelity model. This work presents a damage detection method utilizing Ritz vectors and the Method of Expanded Dynamic Residuals (MEDR). Ritz vectors have several advantages over eigenvectors for application to damage detection, including lower sensitivity to noise, and, as a result of their load-dependent nature, a greater sensitivity to localized damage. The MEDR restricts identified damage locations to those where there is physical connectivity, which eliminates the "smearing" that plagues direct expansion methods, and provides a physically meaningful estimate of the damage location. LA-UR-12-25460. © The Society for Experimental Mechanics 2014.