Towards Autonomous and Robust Damage Detection and Health Monitoring of Mechanical Systems
Date of Completion
Engineering, Computer|Engineering, Electronics and Electrical|Engineering, Mechanical
In this dissertation research, a series of signal processing and decision making techniques have been developed for both machinery condition monitoring (passive detection) and structural damage detection (active interrogation), as one research thrust. In one effort, utilizing accelerometer and microphone signals collected around a gearbox testbed during its operation, a data-driven framework that combines information from multiple sensors and fundamental physics of the gearbox is established to detect gearbox faults passively under non-steady operations (e.g., in wind turbine systems). The feature extraction is facilitated through the usage of harmonic wavelet transform and statistical analysis, and enhanced by incorporating the tachometer readings and gear meshing relationships to develop a speed profile masking technique. In another effort, building upon the Lamb wave based ultrasonic wave approach for structural damage detection, a computationally efficient adaptive harmonic wavelet transform is employed, together with principal component analysis and Hotelling T2 analysis, to elucidate the wave propagation change upon damage occurrence under measurement noise and uncertainty. Analytical, numerical and experimental investigations have been conducted to demonstrate the effectiveness and efficiency of the proposed methodology. ^ A wireless sensing platform is developed, with hardware prototyping and software programming for microphone signal collection/transmission, to demonstrate on-board processing and wireless transmission capabilities. A fundamental hurdle against the wide usage of wireless sensing which presumably is inexpensive and allows dense sensor distribution is the power supply to the wireless sensors. Meanwhile, the ultimate goal of health monitoring and maintenance is to elongate the service life of structures/machineries under adversities such as vibration-induced fatigue/damage. The other research thrust of this dissertation research thus is to develop a new electro-mechanical tailoring scheme using piezoelectric transducers and periodically-arranged circuitry elements to manipulate wave/energy flow. It is identified that, upon proper circuitry tuning, we can create a variety of frequency bands that correspond to different wave/energy propagation patterns, to benefit either vibration-based energy harvesting (to be used in wireless sensing) or vibration isolation (to mitigate high cycle fatigue under tonal vibrations). The components of this dissertation work have laid down a foundation for robust and autonomous damage detection and health monitoring of machineries (e.g., wind turbine gearbox) and structural elements (e.g., wind turbine blades), as well as for self-powering wireless sensors and for concurrent monitoring and vibration control through electro-mechanical tailoring. Together, these components will contribute to the eventual goal of developing sustainable mechanical systems. ^
Lu, Yi, "Towards Autonomous and Robust Damage Detection and Health Monitoring of Mechanical Systems" (2011). Doctoral Dissertations. AAI3504782.