A probabilistic method to assess uncertainty in vision based modal analysis techniques

摘要

This paper addresses the problem of estimating uncertainty when recovering the basic modal parameters of a given mechanical system using a generic vision based displacement measurement system. The particular nature of photogrammetry makes the calculation of uncertainty a challenging task because of the complex phenomena that occur during image formation (motion blur, optical aberration, etc.). To overcome this problem, a Monte Carlo simulation of the whole measurement process (including structure dynamics and its impact on image formation) is proposed. In this way, it is possible to retrieve the posterior probability density functions (PDFs) for the identification of main modal parameters (resonant frequencies and amplitudes). Eventually the proposed probabilistic framework is a suitable tool to assess the performance of a vision based monitoring system on the design stage.