Improvement of Air Fuel Ratio Model Using a Least Absolute Shrinkage and Selection Operator

Abstract

the improvement of an air-fuel ratio model based on in-cylinder pressure measurement is presented. This model has be en applied for calculation of static air-fuel ratio of spark ignition (S I) engines. In addition, it can be replaced the lambda sensors in feedback air-fuel ratio control systems. However the parameters of the model obtained from ridge regression method compose of few small coefficients, in practical applications, these small coefficients cannot be ignored because they affect the airfuel ratio calculation error. Therefore, the authors propose modification of the air-fuel ratio model structure and also the model identification method using a least absolute shrinkage and selection operator (LASSO). With the application of this method, some coefficients are shrunk to zero then the model structure is simplified. The results compared with ridge regression show that this proposed method can improve the structure of the air-fuel ratio model without increasing of the calculation error.