FRF based optimum measurement points selection technique for structural joint stiffness identification

Abstract

This research is focused on the analysis of parameter estimation problem guessing the structural parameters from the measured structural response and system characteristics. Accurate estimation of joint parameters is very important in model updating and diagnosis. Because of the measurement incompleteness and noise,the estimation result is generally inaccurate and highly depends on the experiment condition. The main results of this research are the presentation of optimal sensor placement criterion for the mechanical parameter identification process, and the proposition of a reliable and non-time consuming sensor placement guide. The optimal criteria are derived to get a minimum covariance in parameter estimation process. By using this criteria, an efficient sensor placement guide can be proposed. Unlike most of the other methods, the method in this research does not rely on computationally exhaustive search technique. This method bases on an iterative scheme similar to other methods, which are originated from orthogonal projection scheme and mainly used in state estimation and control processes. In order to overcome a shortcoming of the previous researchers, this method proposes an idea to delete a bunch of degrees of freedom at each iteration stage to get an optimum sensor configuration safely. According to our simulation and experiment results, reliability and rapidity in degree selection process is highly dependent on the choice of information loss at each iteration step. And an efficient parameter identification scheme by minimizing the modal input force is developed to cope with measurement incompleteness and joint uncertainty. The accuracy of proposed method is also dependent on the measurement positions. However proposed measurement position selection guide make the proposed estimation process be robust one. The validity and feasibility of the proposed method are demonstrated both numerically and experimentally. By using this approach parameter estimati...