基于Fluent-BP神经网络的液体动静压轴承热特性分析  

Thermal Characteristics Analysis of Liquid Hybrid Bearing Based on Fluent-BP Neural Networks

作　　者：石莹[1] 王学智[1] 刘金营[2] 付吉海 张校通 

机构地区：[1]东北大学机械工程与自动化学院 [2]95905部队

出　　处：《润滑与密封》2016年第1期37-42,共6页Lubrication Engineering

基　　金：国家自然科学基金项目(51275084);辽宁省重点实验室项目(LZ2015038)

摘　　要：针对液体动静压轴承运转发热复杂的问题,应用Fluent软件对液体动静压轴承进行CFD仿真分析,获得不同输入状态下的油膜温度场分布以及轴承运转时的平均温度和最高温度。并在此基础上通过正交试验将Fluent仿真与BP神经网络相结合,实现对任意输入参数下轴承工作温度的预测,并对转速与供油压力以及供油压力与供油温度的综合作用效果进行分析。结果表明,主轴转速对轴承作用的效果比较显著,当轴承在高转速状态下运行时,需要提供高的供油压力来保证轴承的正常运转;当供油压力下降和供油温度上升同时出现时,轴承运转温度骤升,必须谨慎对待。利用BP神经网络的泛化功能,以少量的样本,可得到均匀全面的网络训练样本点,从而能快捷有效地实现对液体动静压轴承的热特性分析。Aimed at the complex heating problem of liquid hybrid bearing during operation, CFD simulation was used to the rotational speed and oil pressure, oil pressure and oil temperature were analyzed.The results show that the effect of spindle speed on bearing is more significant.When the bearing runs in a high speed,the Oil pressure should be raised to maintain the liquid hybrid bearing operating in a normal state.When oil pressure dropping and oil temperature rising appear at the same time, the operating temperature of liquid hybrid bearing will increase sharply, and it must be treated with the caution. By using the extensive function： of BP neural networks, sample points of network training in full uniform can be obtained with small amount of samples, thereby the thermal characteristics analysis can be implemented quickly and effectively on the liquid hybrid bearing by BP neural networks.

关 键 词：液体动静压轴承 热特性 正交试验 BP神经网络 

分 类 号：TH133.3[机械工程—机械制造及自动化]