Real-time multidirectional modal parameter estimation of beam-shaped objects using high-speed stereo vision

In this study, we develop a dynamics-based visual inspection system that can simultaneously estimate the multidirectional modal parameters of a beam-shaped object under unknown excitation in the audio-frequency range. Our system can measure the three-dimensional (3D) vibration distribution of a beam-shaped object by processing stereo 512×96 images at 10000 fps in real time. Furthermore, it can determine its resonant frequencies and mode shapes along 72 different directions around its axis for inspecting the asymmetric dynamic properties around the axis by implementing a fast output-only modal parameter estimation method on a stereo high-speed vision system. To demonstrate the performance of our system, several steel cantilever beams with artificial cracks vibrating at hundreds of hertz were estimated in real time in order to inspect their asymmetric dynamic properties due to the cracks by estimating their modal parameters along the 72 directions around their beam axes.