AbstractsEngineering

This study aims to develop general methodologies and implementation schemes to enhance the robustness of the time domain methods for damage diagnosis and extend the ability to the assessment of damage severity. Major contributions of this study include 1) the development of a novel time-series analysis method, which requires only acceleration response signals from the structure, for detection of the occurrence and location of damage; 2) extension of the above method to noise-contaminated vibration signals, with a novel scheme incorporating Kalman filter to establish the virtual Input-Output signal pairs that essentially represent the underlying physical system; and 3) formulation of a residual generator technique, based on geometric concept for disturbances decoupling problem (DDP), for detecting and locating the damage using acceleration measurements instead of displacements. The method also enables quantitative estimation of the severity of damage in individual elements of a complicated system. Numerical examples and experimental case studies are given to demonstrate the implementation and effectiveness of the proposed approaches.