内燃机机油泵性能建模方法的比较研究  

Comparative Study on the Model Methods of Performance of Oil Pump in Engines

作　　者：童宝宏[1] 桂长林[2] 何心胜[1] 张强[1] 任思路[1] 施超[1] 陈海兵[1] 

机构地区：[1]安徽工业大学机械工程学院,安徽马鞍山243002 [2]合肥工业大学机械与汽车工程学院,安徽合肥230009

出　　处：《内燃机》2009年第5期11-13,共3页Internal Combustion Engines

基　　金：国家自然科学基金资助项目(50575065);安徽省高校优秀青年人才基金资助项目(2009SQRZ077);安徽工业大学大学生科研训练计划资助项目(08043)

摘　　要：通过对目前几种常用的机油泵工作特性建模方法进行比较分析,指出全工况下对机油泵性能进行建模的必要性。结合前期研究结果,对BP网络、延迟BP网络、径向基函数网络和反馈型神经网络这四种典型的神经网络模型进行实例比较研究,从而得到预测全工况下机油泵性能的最佳神经网络模型。还对神经网络建模方法的不足以及相应的联合建模方案进行了初步探讨,提出利用先进的CFD仿真分析手段,结合尽可能少的实验数据,实现全工况下的机油泵性能的预测和建模。A comparative study on the model methods of performance of oil pump in engines is introduced and it points out that modeling for oil pump＇s performance under various work conditions is necessary. Combining with early researching, comparisons among BP neural network, delayed BP neural network, RBF neural network and elman neural work are also introduced by an example study. And the best model method is reached to predict the pump＇s performance under various work conditions. Defaults of neural network are pointed out, and a co-modeling method is discussed. It promotes that modeling and prediction of oil pump＇s performance can be realized combining with advanced CFD simulation method and least experiments.

关 键 词：内燃机 机油泵 建模 比较研究 

分 类 号：TK401[动力工程及工程热物理—动力机械及工程]