Identification Of Unmodelled Dynamics In A Rotor-Bearing System With Bounded Uncertainty

In developing engineering models for rotating machine analysis, most of the components of the rotor–bearing– foundation system can be modelled relatively accurately. However, there are always some parts or interactions which are very hard to model, such as shrink fits or foundation components. Such parts or effects can be referred to as unmodelled dynamics. This paper presents a non-parametric method to identify the unmodelled dynamics using the combination of an engineering model and experimental data. Assuming that errors in the engineering model and measurement noises are bounded, a bound is derived for the uncertainties in the identified dynamics by using µ-analysis. An illustrative example is given, and two widely used non-linear searching methods are compared to the µ-analysis in estimating the bound, showing that the µ approach is the only efficient method to find a tight hard bound.