滑动轴承非线性油膜力的神经网络模型

在已有的滑动轴承非线性油膜力数据库基础上 ,将轴承的位置和速度参数加以综合 ,利用变量状态空间变换将分段的油膜力数据转换成连续的数据空间 ,建立非线性油膜力连续型数据库和相应的网络模型 .以圆轴承 -转子系统为例 ,分别采用有限差分法、数据库法和 BP网络模型计算了轴承系统的非线性油膜力和轴心轨迹 .结果表明 ,网络模型计算结果与基于数值方法的结果较为吻合 ,可以显著地提高轴承系统的计算效率

BP Network Model for Nonlinear Oil-film Force on Hydrodynamic Bearing

The database method to compute nonlinear oil film force of the finite width hydrodynamic journal bearings can solve the contradiction between accuracy and efficiency, but the database and formulae of oil film force are subsections. Positions and velocities of bearings are synthesized into three basic parameters and state space transformation is used to change discontinuous oil film force databases to consecutive. So the integrative formula and BP network model based on consecutive databases forces are established to obtain oil film force under any movement states of axes with high accuracy. By means of computation example of cylinder journal hydrodynamic bearings, finite differential method, database method and BP neural network models of nonlinear oil film forces are employed to calculate oil film forces and orbit of shaft center in the transient analysis of the rotor bearing system. Results are shown that BP network model is more approximate to numerical computation methods and raises remarkably computational efficiency of bearings.