Date of Graduation


Document Type



This dissertation presents a non-destructive automated damage detection system for the Armored Vehicle Launched Bridge. The goal of the system is to automate the damage detection procedure and enable early detection of structural damage. The system would complement the existing visual inspection procedure and eventually replace it. The designed system utilizes the method of strain energy mode shapes as the core damage detection method. The presented research covers both practical design issues and new theoretical developments in the area of non-destructive damage detection using strain energy mode shapes. The dissertation presents a methodology for exciting the structure, acquiring the vibration response of the structure, validating the acquired data, determining vibration characteristics of the bridge (such as natural frequencies), extracting the displacement mode shapes, applying the strain energy processing to the displacement mode shapes, and analyzing the strain energy mode shapes for damage indicators. The system employs a laser Doppler vibrometer mounted on a robotic gantry crane as the vibration sensor. Two testing techniques (sinusoidal dwell testing and random burst testing) are supported by the system. The system is capable of extracting damage indicators through a traditional formulation of strain energy method and a novel non-baseline variation of the strain energy method that does not require a-priori knowledge of the undamaged state of the structure. A fuzzy expert system enables the system to detect and locate damage on noisy data with the performance comparable to those of a qualified human operator. The test procedure is fully automated by a custom-designed application software. Investigated theoretical issues present new developments in the area of damage detection using strain energy mode shapes. An analytical study of the numerical properties of the strain energy processing allows proper selection of the sampling interval during acquisition of the displacement mode shapes, minimizing the effects of measurement noise and providing maximum sensitivity to damage, while improving the accuracy of damage location. A novel approach to the extraction of damage indicators from the strain energy mode shapes has been proven analytically and verified experimentally. The new approach significantly extends the applicability of damage detection using strain energy mode shapes to a structure with previously unknown damage state.