The Mechanism Analysis And Intelligent Diagnosis Of Coupling Faults Based On Rotor-Bearing Coupling Dynamic Model

Rubbing of the rotary machine's rotor system is usually the second fault which results from unbalance, misalignment and oil whirl and so on. The signal output usually has three characteristics: periodic, quasi-periodic, and chaos, it has complicated nonlinear characteristics. The general high speed rotary machines are always supported on dynamic pressure sliding bearing. But for some rotary machines, of which the weight and structures are highly exigent, such as aero-engine, usually adopt rolling bearing. The two types of bearing are applied widely.Firstly, in this paper the rotor- sliding bearing coupling system model with coupling faults is established, in the model the nonlinear oil whirl force is considered, three faults, namely, mass unbalance, oil whirl and rubbing are included. Secondly, the rotor-rolling bearing coupling model is established, in which the Hertzian nonlinear contact and the radial bearing clearance are considered. Finally the rubbing mechanism of the two rotor systems is studied, the frequency domain characteristics are extracted, and the characteristics information of the couple faults are analyzed by using the orbit of shaft center, frequency spectrum, three-dimensional frequency map, poincarémap, bifurcation plot and so on.After enough rubbing samples in various states are obtained through numerical simulation of two dynamic models, a structure self-adaptive neural network is constructed, the simulation samples are used to train and test the net in order to obtain the most excellent net configuration. Finally, the successfully trained network is used to diagnose rubbing coupling fault of samples from multi-functions rotor simulation testing desk and aero-engine rotor testor. The results validated the usefulness of the dynamics models which are established in this paper.Finally, in order to extract the coupling faults characteristics further, the correlation dimension theory is applied to the rotor systems, used the nonlinear time series analysis theory, through the phase space reconstruction of the signals obtained from the dynamic models, the correlation dimension is calculated. The discipline of the correlation dimension of coupling faults is studied and is acted as the important characteristics to identify the coupling faults.