A Model-Based Identification Approach for Local Fault of Rotating Machinery

Model-based identification technique is now widely used in condition monitoring and fault diagnosis for rotating machinery. With the help of it, the fault position and transient fault force could be identified for many local faults, including weak rubbing, transverse crack, coupling misalignment, and pedestal looseness. In this paper, the model-based identifying theory and its main procedure are studied. At first, a hybrid model combining a finite element model of rotor and rigid discs with online identified oil-film coefficients is adopted. Modal expansion and least square fitting is proposed to locate for fault position. The transient fault forces are then identified from transient residual vibration based on matrix operations. These approaches are successfully applied to the identification of the local rubbing fault occured on a simple rotor test rig with two discs.