Estimation of instantaneous states of an SI gasoline engine using EKF and UKF

This paper deals with the estimation of instantaneous states of a Spark Ignition gasoline engine which is a hybrid system exhibiting both continuous and discrete dynamics. Two estimation techniques have been explored, namely the Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) to operate on such a model. It has been shown in this paper that the UKF estimator performs better than the EKF under same model and measurement conditions. Hence proving the fact that the UKF is better for estimation of highly non-linear switched dynamic system like engines as it does not approximate the model by linearization, by the computation of Jacobians, which is done in case of EKF. The estimation is done for 12 states of a 4-cylinder engine with 5 measurements and results have been validated using standard engine simulation software.