Model-based interpolation-iteration method for bearing coefficients identification of operating flexible rotor-bearing system

Abstract

The requirement of bearing coefficients identification based on flexible rotor-bearing systems has stimulated the investigation of bearing forces over the years, whereby in industry high speed balancing machines and flexible rotor-bearing test rigs can be used to measure the dynamic force coefficients of bearings. However, the actual rotor vibration at the bearing nodes cannot be measured or simply calculated by the data acquired from a single measurement station outside of the bearing. To solve this problem, a double-section interpolation-iteration method to identify the dynamic coefficients has been developed which uses previously identified coefficients to predict new vibration vectors from a finite element model and then updates the vibration vectors at the bearings to recalculate the coefficients. Numerical investigations on the estimation error, convergence, and robustness against noise were carried out, and the specific measurement process and method are presented. Simulation results show that this method has high confidence intervals in the identification of both direct stiffness and damping coefficients. Moreover, the experimental work, in which an active magnetic bearing was used to excite a flexible test rig supported by tilting pad bearings, was also carried out to validate the method, with the vibration shape predicted by the identified results matching the test data very well and the first forward mode parameters also matching the results from sine-swept identification.