Analysis of nonlinear vibrations and health assessment of a bearing-rotor with rub-impact based on a data-driven approach

Affected by manufacture, assembly, and operating conditions, the rotor is prone to rub-impact fault, resulting in instability of a system. The nonlinearity of a system response includes instability and fault characteristics. Based on this, this paper proposes a novel health assessment approach for bearing-rotor system that combines data-driven and NOFRFs (nonlinear output frequency response functions), revealing the intrinsic relationship between NOFRFs-based feature and rub-impact fault. Before utilizing the approach, the vibration and stability of the bearing-rotor with rub-impact are simulated and analyzed. Based on Hertz contact theory, the nonlinear bearing forces are obtained. Then the dynamic model of the rolling bearing-rotor system considering unbalance and rub-impact is established. The influence of rub-impact on the nonlinear vibration and stability of the bearing-rotor system is analyzed by combining the orbit, the Poincaré map, and the bifurcation diagram. A rolling bearing-rotor test rig is built, and the rub-impact experiment was completed. The data-driven model of the system is identified by the filtered vibration displacement. The NOFRF-based feature of the data-driven model is extracted. With the increase of rub-impact severity, the dispersion of feature gradually increases, and the stability of the system decreases. The standard deviation of the feature is solved as a new health indicator. The result shows it is more effective for the evaluation of rub-impact than the conventional indicators.