Study on tribo-dynamic behaviors of rolling bearing-rotor system based on neural network

Abstract The tribo-dynamic model of the rolling bearing-rotor (RBR) system based on the radial basis function neural network (RBFNN) is proposed to effectively obtain tribo-dynamic performances of the system. In doing so, the thermal elastohydrodynamic lubrication (TEHL) and motion equations of the rolling bearing are solved simultaneously for the oil film forces, and the fast Fourier transform (FFT) is employed to accelerate the deformation computation. Further, the RBFNN is well trained to reconstruct the oil film forces of the bearing. By incorporating the oil film forces into the dynamic simulation module of the RBR system, its transient tribological and dynamic performances are predicted using the Matlab/Simulink module. Meanwhile, the efficiency and accuracy of the RBFNN in predicting the oil film forces are verified. Then, the tribo-dynamic performances of the RBR system such as the spiral structures of the film thickness, film pressure and film temperature are revealed.