A non linear observer for fuel film dynamics into the intake manifold of a spark ignition engine

A nonlinear observer, based on an extended Kalman filter, is presented for the estimation of the nonlinear fuel film dynamics into the intake manifold of a spark ignition engine. In order to meet the stringent limits imposed on automotive exhaust emissions, electronic engine control systems have to constrain air-fuel ratio in a narrow band around the stoichiometric value, since even a small excursion results in a large degradation of catalyst conversion efficiency. A two states dynamic model is used to describe liquid and vapour fuel mass flows into the intake manifold, accounting for both the partial impingement of the injected liquid fuel on the manifold walls and the evaporation process. The application of an extended Kalman filter allows to deduce a minimum error estimate of both model states (liquid and vapour fuel mass) and parameters (evaporation time constant and injected liquid fraction impinging the wall) utilizing the knowledge of system dynamics and the UEGO sensor measurement at exhaust location. The observer has been tested on a set of 35 air-fuel ratio dynamic transients showing a good prediction level with respect to engine fuel flow and air-fuel ratio. Moreover the trend of estimated model parameters vs. operating conditions is consistent with the physical phenomena taking place in the region interested by the fuel injection process