The dynamic characteristic analysis and neural network control of a spiral bevel gear transmission system

A parametric solution domain structure based on the cell mapping theory and a control method based on a radial basis function neural network were proposed to solve the dynamic characteristic transition and the control problem of a bevel gear system with backlash. A dynamics model of 7-degree-of-freedom spiral bevel gear was established by using the concentrated mass method. Then, the frequency and load parameter plane was constructed based on the cell mapping theory, and the pseudo-fixed point continuous tracking algorithm was used to solve the transition rule of bifurcation, tooth surface impact, tooth no-meshing, tooth back contact and dynamic load characteristics of the straight bevel gear system. It was found that frequency and tooth impact are the main factors affecting the periodic bifurcation. With the increase of load, the tooth no-meshing and impact weaken, and the dynamic load coefficient increases. For the chaotic phenomena of the system in the plane, a parametric feedback controller was designed, and the fitness function was constructed based on Poincare cross section Euclidean distance. The adaptive gravity search algorithm was used to optimize the controller parameters, so as to realize the effective control of chaos, quasi-period and periodic motions to periodic orbit. © 2024 Chinese Vibration Engineering Society. All rights reserved.