Identification of speed-dependent bearing parameters

The reliable estimation of a flexible foundation model and the state of unbalance (both amplitude and phase) of a turbogenerator from machine run-down measured vibration data is an active research area. Earlier studies on the estimation of both these quantities used the whole frequency range of the run-down as a single band. However, such an identification may be inaccurate for large flexible foundations having many modes in the run-down frequency range. For reliable identification, the whole frequency range has to be divided into a number of frequency bands and the frequency dependent foundation models have to be estimated together with the unbalance. This paper combines the unbalance estimation with the split frequency range for the foundation model, and the highlights the limitations observed during estimation of foundation models and the state of unbalance. A simulated example is used to validate and assess the advantages of the proposed method. This method is compared to the previous method using the whole frequency range as a single band. Having established the method in simulation, experimental data from a 3 m long test rig, with four journal bearings, is used to test the method. The approach gives reliable estimates of the machine unbalance, even in the presence of modelling errors and measurement noise.