Impact of measurement errors in inverse analysis

During the first step in the development of any structural system identification method, the method should be validated by noise-free measurements in the first place. Nevertheless, this assumption is far from reality as the measurements in these tests are always subjected to the errors of measurement devices. To fill this gap, this paper analyzes the effects of measurement errors in a parametric structural system identification method: the observability method. To illustrate the symbolic approach of this method a simply supported beam is first analyzed in detail. This simulation provides the parametric equations of the estimates. Then, the effects of errors in a particular measurement, errors in all measurements, load locations are studied in this structure. Two additional examples of increasing complexity are also analyzed to show the effect of modelling errors on the estimates. A fluctuation of the observed parameters around the real values is proved a characteristic of this method. The results of these structures illustrate how important the effects of modelling errors are especially in areas with low curvatures.