Fast analysis of motion state for elliptic bearing-rotor system

Based on continuity analysis of oil-film force databases of hydrodynamic bearings, a neural network model of nonlinear oil-film forces of elliptical bearings is developed. The neural network is trained by chaotic BP algorithm. The solutions of network model are almost identical to numerical computation, and the running efficiency is increased compared with the database computation method. The motion of elliptical bearing-rotor system is simulated by means of network model of oil-film forces and fourth-order Runge-Kutta method. Nonlinear motion states of bearing-rotor system are obtained with high efficiency and accuracy. There exist multi-types of motion behaviors at various rotating speeds, such as synchronous, subsynchronous, quasi-periodic or chaotic motion. It is shown that the neural network models of oil-film forces are useful to study nonlinear dynamic problems of rotation machinery.