基于NARX的SI发动机空燃比非线性模型预测控制

针对SI(Spark Ignition)发动机空燃比(AFR: Air-Fuel Ratio)控制精度低、无法自适应等问题, 提出了基于NARX(Nonlinear Auto Regressive model with eXogenous inputs) 模型的非线性模型预测控制(NMPC: Nonlinear Model Predict Control)方法。利用渐消记忆递推最小二乘(RLS: Recursive Least Squares)算法对NARX 模型进行辨识, 基于NARX 模型对SI 发动机的AFR 进行非线性模型预测控制。该方法辨识精度高, 可通过NARX 模型数学结构直接计算最优控制序列, 从而提高系统的控制精度。同时, 采用Matlab 对均值发动机模型(MVEM:Mean Value Engine Model)进行仿真实验, 并与采用Volterra 模型的PI(Proportional Integral)控制器算法进行对比。仿真结果证明, 该算法控制效果比基于Volterra 模型和传统的PI 控制器的控制效果超调量小, 调节时间短, 更加具有工程实际应用性。

NARX Model Based Nonlinear Model Predict Control for Air-Fuel Ratio of SI Engines

For SI( Spark Ignition)engine AFR( Air-Fuel Ratio)control system,the NMPC( Nonlinear Model Predict Control)method based on the NARX( Nonlinear Auto Regressive model with eXogenous inputs)model is developed. First the algorithm of recursive least squares is used to identify the NARX model. And then use the method of non-linear model to predict control in SI engine. This method has high prediction accuracy,and the optimal control sequence. Thus it improves the SI engine AFR control performance. Finally,the control system is evaluated using a widely used SI engine benchmark model which is Mean engine model by using Matlab,and compared with the PI( Proportional Integral)controller algorithm which current widespread use of car and Volterra model. Simulation results demonstrate the effectiveness of the NARX model based nonlinear model predict control.