基于BP神经网络的发动机转矩估计  

Engine Torque Estimation Based on BP Neural Networks

作　　者：杜常清[1] 颜伏伍[1] 杨平龙 杜传进[1] 

机构地区：[1]武汉理工大学汽车工程学院,武汉430070 [2]长安伟世通发动机控制系统有限公司,重庆401122

出　　处：《汽车工程》2008年第7期588-591,共4页Automotive Engineering

基　　金：国家863计划项目(2005AA501220-3)资助。

摘　　要：针对并联式混合动力总成在状态切换和换挡时动力协调控制的需要,完成了富康TU5 JP/K汽油发动机的稳态和动态性能试验,建立了基于BP神经网络的发动机转矩估计模型:通过对试验数据的训练,完成了全工况范围内的发动机稳态和动态转矩的估计,并实际应用到在基于dSPACE的混合动力总成控制原型中,达到了预期的控制精度,为实现并联式混合动力系统在状态切换和换挡过程中发动机和电动机的协调控制奠定了基础。In view of the requirements of coordinated control for parallel hybrid powertrain system during mode switching and speed shifting, the steady and transient performance tests of Fukang TU5JP/K gasoline engine are conducted, and a BP neural network-based model for engine torque estimation is set up; Through the training on test data, the estimation of both steady and transient engine torques in whole range of working conditions is completed, which is then practically applied to the dSPACE-based control prototype for hybrid powertrain with expected control accuracy attained. These lay the foundation for realizing coordinated control of engine and electric motors of parallel hybrid powertrain system in mode switching and speed shifting processes.

关 键 词：车用发动机 BP神经网络 转矩估计 

分 类 号：U464[机械工程—车辆工程]