基于模型在环仿真的直喷汽油机空燃比辨识与控制研究  

Model-Based Identification and Control of GDI Air-Fuel Ratio

作　　者：胡春明[1] 胡东宁[1] 刘娜[1] 

机构地区：[1]天津大学内燃机研究所,天津300072

出　　处：《内燃机工程》2016年第3期88-93,共6页Chinese Internal Combustion Engine Engineering

基　　金：国家自然科学基金面上项目(51476112);天津市科委科技支撑计划项目(12ZCZDGX03800)

摘　　要：针对过渡工况下汽油机空燃比难以精确控制的特点,将PID神经元网络(proportional-integral-derivative neural network,PIDNN)控制策略用于发动机常用的部分负荷工况下空燃比的辨识与控制。在AMESim中,从实际发动机物理模型出发建立了发动机仿真模型;同时在Matlab软件中建立PIDNN辨识器、PIDNN控制器;通过Matlab和AMESim软件的耦合进行模型在环仿真,建立电喷发动机实时控制系统,通过联合仿真观察辨识效果,同时检验控制系统的性能。模型在环仿真结果表明:PIDNN辨识器能够快速跟踪实际空燃比的变化,辨识系统的输出与发动机模型的输出误差在0.05之内;在过渡工况下,PIDNN控制器控制下的过量空气系数超调量相对于普通PID控制器能减小0.3以上,同时能在2s内将过量空气系数控制至目标值,提高了空燃比的控制精度。Since the air-fuel ratio of gasoline engines is difficult to control accurately under transient conditions, the PID neural network was applied to identify and control the partial load air-fuel ratio. An engine model was built in AMESim according to a real engine parameters. Meanwhile the engine controller was built in the Matlab/Simulink software. Using the interface between two softwares , the co-simulation was made to identify and control the air-fuel ratio in real time. The simulation results show that the identification system can quickly tracked the real air-fuel ratio with a deviation of less than 0.05. The controller can effectively control the air-fuel ratio with lower overshoot about 0.3 than that of the normal PID controller, and quick response with a normal transient period of less than 2 s, which shows the validity of the control of the air-fuel ratio.

关 键 词：内燃机 PID神经元网络 过渡工况空燃比控制 系统辨识 模型在环仿真 

分 类 号：TK411.12[动力工程及工程热物理—动力机械及工程]