Model Acquisition for Modal Analysis of Flexible Media Based on Stereo Vision

Abstract

To enhance and ensure machine reliability and accuracy, obtaining its mechanical properties in real-time dynamic conditions is of engineering importance. In both roll to roll printing and metal sheet rolling processes, it is well known that any misalignment or orientation arrangement between rollers can significant change stress distribution in the high speed transporting flexible media, leading to quite different dynamic responses in the manufacturing process. In sheet metal manufacturing process, to eliminate residual in the flexible media, it is a common process to place two adjacent rollers at different orientations and location, causing the media to deform largely and even to twist. In this case, common non-contact vibration measurement method using either single point or scanning laser Doppler vibrometry cannot accurately capture the vibratory response from the media due to the significant change in geometry to scatter the reflective laser light. In this paper, instead of adjusting the laser perpendicular to the curved surface of the media at each measurement point to capture modal information through experimental modal analysis, which is achieved by utilizing the excitation and measurement apparatuses to directly discover the modes of the media, a novel non-contact measurement method is proposed by using stereo vision to first acquire the shape of the flexible homogeneous metal sheet through one camera with a fixed focal length from two different views, and then calculate the coordinate by computer vision algorithm. The calculated coordinates and shape information will be input to the commercial FEA software and compute its modal information, namely natural frequencies and mode shapes. Lastly, the results gained by the method proposed will be compared with that acquired directly by the experimental methods.