SI发动机空燃比联合非线性模型预测控制

Joint non-linear model predict control for air-fuel ratio of SI engine

针对火花塞点火(SI)发动机空燃比(AFR)控制系统提出了一种变采样间隔的Modified Volterra模型,并以此为基础,提出了一种基于RBFNN和Modified Volterra模型的SI发动机AFR的联合NMPC控制方法。该方法既具有RBFNN模型计算量小、预测精度高的特点,又具有可直接计算NMPC最优控制序列的优势,明显地提高了SI发动机AFR的控制精度,大大地减少了常规迭代寻优算法的计算时间。在dSPACE实时仿真试验平台上对平均值发动机模型进行仿真试验,结果表明:本文所提出的NMPC控制方法对SI发动机AFR的控制效果明显优于单独基于Modified Volterra和RBFNN模型的NMPC控制方法。

A variable sampling period Modified Volterra model is proposed for the Air-Fuel Ratio （AFR） control system in Spark Ignition （SI） engine. On this basis, a joint Nonlinear Model Predictive Control （NMPC） method is developed based on the Radial Basis Function Neural Network （RBFNN） model combining with the modified Volterra model. The advantages of this method are small amount of calculation and high prediction accuracy; also the optimal control sequence can be directly calculated. Thus it significantly improves the AFR control performance of SI engine, and greatly reduces the computing time compared with the conventional iterative optimization algorithm. Real-time simulations based on the mean value engine model are conducted on the dSPASCE simulation platform. Results show that the control performance of the proposed method is significantly better than the RBFNN model or the Modified Volterra model based NMPC method.