Iterative learning-based air-fuel control of gasoline engines with unknown off-set

This paper presents an air-fuel ratio control scheme via individual fuel injection for gasoline engines with multi-cylinders. The control scheme targets on unknown off-set in the fuel path of individual cylinder which represents the effect of the cylinder-to-cylinder imbalance caused by the perturbations in each injector gain or other disturbances in the dynamics of fuel injection path. The fueling-to-exhaust gas mixing system is treated as single-input single-output (SISO) periodic time-varying system where the input is fuel injection command for each cylinder and the output is the exhaust bottom dead center (BDC) scaled air-fuel ratio sampling. Then, an observer-based iterative learning strategy is proposed to provide estimation of the unknown off-set in each cylinder. The effectiveness of the learning algorithm will be demonstrated with the experiment results conducted on a commercial car-used engine with six cylinders. Furthermore, as an application of the learning algorithm, feedback control with feedforward compensation is also demonstrated with experiment validation.