Comparison of the Extended Kalman Filter and the unscented Kalman filter for parameter estimation in combustion engines

Abstract

In this article, two Kalman filtering techniques, the Un-scented Kalman Filter (UKF) and the Extended Kalman Filter (EKF) are applied for cylinder-wise torque estimation. In engine signal processing the problem of engine speed evaluation is one of the main problems of current research for engine control. In this work, two engine speed signals, recorded at the free end and at the flywheel, together with a multi-body model of the crankshaft are used to account for torsional deflections of the crankshaft. In order to estimate cylinder-wise torque, additionally one cylinder pressure signal is used to obtain a parametric torque model. The resulting parameter and state estimation problem allows the comparison of UKF and EKF. The performance of both algorithms was evaluated using measurements from a four cylinder combustion engine. Whilst practical issues still exist, this off-line study showed the feasibility of the approach.