Identification of unbalance in a rotor-bearing system using Kalman filter-based input estimation technique

Unbalance is a common machinery fault, which occurs because of uneven distribution of mass of a rotating component about its axis. Vibration-based monitoring techniques have been widely accepted for machinery fault diagnosis. This article presents experimental verification of a recently developed Kalman filter-based method for the identification of unbalance in rotor systems. The method is tested on an experimental test rig for different unbalance configurations and shaft speeds. The proposed technique is a model-based method, which requires a mathematical model of the rotor system along with response measurements. A rigid rotor model is used, and measured accelerations at bearing pedestals are used for unbalance parameter estimation. Bearing stiffnesses are estimated using a frequency domain parameter estimation technique with measured unbalance responses. Sensitivity analysis is also performed by altering the values of these estimated stiffnesses.