| This work addresses the problem of incipient fault detection and diagnostics for rotating machinery. This includes problem description, modeling of rotating machinery, fault mechanisms, and a fault diagnosis scheme. The proposed scheme is implemented in a simulation environment to test its feasibility. In this work, the faults (i.e. rub) of rotating machines are modeled in terms of nonlinear dynamic systems with quasi-periodic and chaotic behavior. It is shown that the non-smooth dynamic model of rub impact of rotating machines poses technical challenges to diagnosis algorithms that are based on nonlinear observers (such as, extended Kalman filter “EKF”). To avoid those challenges a regression viewpoint of the model was adopted, resulting in robust and effective algorithms for the monitoring and diagnosis of the rotor-stator process.; The regression-model approach is founded on computationally efficient algorithms for signal processing and parameter identification. A simulation study, which includes normal and different fault modes, illustrates the performance of the proposed approach, especially in the presence of measurement noise and process uncertainty.; Because of its robustness and simplicity, the proposed model-based approach is potentially of considerable value as a diagnostic tool in assessing condition-monitoring signals that are now routinely taken on modern rotating machinery. |
