Air-Fuel Ratio Control For SI Engine Based On MATLAB/RTW

To control the AFR of SI engine in high precision is an important measure to reduce the emission of automobile and the deterioration of the air. Aiming to the subject of AFR accurate controlling of gasoline engine fuel injecting, The author builds the perfected mean value model, applies RBF(radial basis function) Neural Network and Adaptive Inverse theory to accurately control the AFR based on the model for SI engine in this paper.First, the author begins with the necessity of control for AFR (especially the transient AFR) in an automobile and researches on this issues that have done both domestic and overseas. Mean-while, it puts forwards its own mean value models of SI engine based on the synthetically analyzing of engine models that have existed till now. It develops a no algebraic loop sub-model for TFC, thus, it perfects the mean value model of SI engine based on Simulink.Second, the author adopts the NN and nonlinear theory to control the transient AFR based on analyzing the existed transient AFR control strategy. After comparing the advantage of RBF with any other NN, choose the RBF NN to control the transient AFR. The algorithm of RBF has many kinds, choose the maxim matrix element to determine the number of hidden layer and make use of an improved algorithm based on Gaussian Kernel Function to initiate parameter and its learning method in this paper. Subsequently, the author finish the online simulation of RBF AFR control, the result show that the controller has good adaptation and achieve accurately control AFR. Takes advantage of integral design for control system based on MATLAB/RTW, separately accomplishes the generic real-time target code generation for the AFR control system and the single controller. The simulation verifies that the code is valid.Third, after introducing the theory of nonlinear Adaptive Inverse control and contrasting to all kinds of adaptive algorithm, adopts the Adaptive Inverse control strategy based on an improved LMS algorithm. Separately fish the simulation with M s-function and C-MEX s-function controller based on Simulink. The result show that the both controllers achieve accurate AFR control with the only±1% error at complex condition, the C-MEX S-function controller accelerates the simulation. This paper accomplishes the generic real-time target code generation for the C-MEX S-function controller with continuous and dispersed states .The engine has many indices, the main indices are economic and power. The different engine work conditions have different demand mix gas. To achieve the full condition AFR control, make use of the finite state machine theory to build the Stateflow chart for engine based on the relation among the different engine condition. The simulation shows that finite state machine theory is suit for disposing of the finite complex states. The automatic generation of C code from Stateflow chart is efficient and the code is suit for making the Simulink/Stateflow model to be the HLA member by future optimizing. In the end, the achievements and some problems in the research are summarized in the dissertation, and some suggestions for future study are provided.