Control-oriented Modeling of a Turbo-charged Gasoline Engine by Fusion of Physical Modeling and System Identification

Abstract

Turbo-charged engine is a system contained complicated non-linear characteristics. In addition, the system have constrains for actuators and internal states. Therefore, to apply a nonlinear control system design such as Nonlinear Model Predictive Control (NMPC) is highly expected. In this paper, we propose a control-oriented model of the engine by fusion of a physical modeling and a system identification. The proposed model is derived by three steps. For the first step, we construct a control-oriented reduction model using Lasso regression analysis method for each element system of the engine. The second step, we combine the element models to give a state-space model of the entire turbocharged engine. The third step, we construct an error compensation model to improve the accuracy of the state-space model. The effectiveness of the proposed method is shown by using experiments data from actual engine.