Multi-parameter Model Identification Of Air-Fuel Ratio And MPC Controller Design Of Gasoline Engine

Most of traditional gasoline engines adopt port fuel injection, so there will besufficient time for air and fuel to mix. Then the combustion and the efficient of theengine will be better after the homogeneous charge go into the combustion chamberand start burning. However, this method is easier to form fuel film on the intake portor the inlet valve, thus affecting the precise control of engine air-fuel ratio. But today,the post-processing technology is widely used, the high efficiency of thepost-processing device is dependent on the precise air-fuel ratio control. During coldstart of engine, the film phenomenon is more obvious and there will be more time lag.So the precise control of the air-fuel ratio will be faced greater challenge.To solve this problem, the traditional method is to build the physical model ofinlet fuel film which is to describe the relationship between the injection rate and theair-fuel ratio. In steady state, the air fuel ratio control in which the identified model isutilized have a good performance. But the defect is that when the engine is operated intransient conditions, especially in the cold start and slow down conditions, theperformance of the model is not so well. At the same time, it’s so simple of the singleinput and single output model that it’s not concerned the other parameters which areaffected the fuel film and air-fuel ratio. So the N4SID method which is one of thesubspace methods is used to identify the multi-parameters model of air-fuel ratio. Andthen, the control of the air-fuel ratio is studied. The results have shown that it issimple enough to use the N4SID method. The accuracy and the anti-interferenceability of the model are well. To further improve the accuracy of the model, thepredictive error method (PEM) is used to optimize the identified model. Thesimulation results show that the optimized model is more robust and accurate. Themodel performance of the convergence is improved.Based on the N4SID model, the multi-parameters of model predictive controller(MPC) is designed. The detail design process of the controller is studied. Thesimulation results show that the step response of the controller is stable and thefluctuation of the predictive control signals are small, so the controller is adjusted toapplication.