Adaptive in-cylinder pressure model for spark ignition engine control

Abstract

In-cylinder pressure is the most important variable to analyze the combustion process in internal combustion engines, and can be used as feedback signal for closed-loop combustion control and diagnostics. However, pressure sensors are still affected by challenges such as durability and cost, which prevent their use in mass-production vehicles. Therefore, this work presents a model-based approach to estimate the in-cylinder pressure by means of the combination of a control oriented model and information from the set of sensors available in current production engines for automotive applications. Pressure peak location of each cylinder is estimated through the knock sensor signal, and used as feedback to improve the model. An extended Kalman filter is used to adapt the model to the information from the knock sensor signal. The adaptive model is implemented in a four cylinder light-duty engine and compared with the open loop model, ensuring a continuous estimation of in-cylinder pressure signal, and an improvement for the estimation of different cycle by cycle combustion parameters and cylinder to cylinder variations. Finally, the proposed approach is applied with different fuels showing that the proposed method can be applied independently on the fuel used.