Diesel engines have widely applied to passenger cars along with significant technological developments to reduce nitrogen-oxides (NOx) and particulate matter (PM) emissions. The representative components to reduce engine out emissions and improve fuel efficiency are exhaust gas recirculation (EGR) system, the variable geometry turbocharger (VGT), and the common rail direct injection (CRDI) system. Among them, EGR and VGT systems for the control of the diesel engine air system have the nonlinear multi-input-multi-output (MIMO) characteristics. These nonlinear characteristics cause difficulties in precise control. In addition, the reference variable, mass air flow (MAF), for the control of the EGR system has weak correlation with NOx formation, and the air system control without emission feedback leads to drift of engine out emissions in accordance with the engine operating conditions.
This thesis investigates the nonlinear model-based control of the EGR and VGT systems for a passenger car diesel engine using the intake oxygen concentration (IOC) and manifold absolute pressure (MAP) with the emission feedback for the compensation of NOx drift. The IOC has strong correlation with the formation of NOx emission. Thus, the IOC is used as a reference variable for the control of the EGR system.
Additional informations about the engine those cannot be measured in mass production engine are required for the development of model-based control algorithm of the EGR and VGT systems. The nonlinear mean-value air system model is used to derive unmeasurable states of the target engine. The air system model consists of four thermodynamic and seven empirical sub-models. The IOC which is used as a reference variable for the EGR control is estimated from the closed-loop observer, and the robustness of the observer is validated using the input-to-state stability theorem.
The model-based control algorithm of the EGR and VGT systems generates the feedforward values of the EGR valve lift and VGT vane positions to track the set-values of the IOC and MAP with consideration of the engine operating conditions. The EGR control algorithm consists of 2 nonlinear mean-value models and the VGT control algorithm consists of 4 nonlinear mean-value models. The modeling error of the model-based feedforward control algorithm is compensated with PI control algorithm. The experimental results of changing the IOC set-value with the model-based feedforward control algorithm represent improved transient responses; the rising time is improved more than 60% compare with the look-up table based feedforward control algorithm.
The emission drift in accordance with engine operating conditions is addressed with emission feedback control. The emission feedback control focuses on the reduction of increased NOx emission due to the NOx drift. It is assumed that the PM emission is entirely manageable by the PM after treatment systems. The engine out NOx is measured using the on-board NOx sensor. The emission feedback algorithm corrects the IOC set-value, which has strong correlation with the formation of NOx emission, when the measured NOx exceeds reference NOx level.
The feasibilities of the proposed mean-value air system model, IOC observer, model-based control algorithm, and emission feedback structure were validated with engine experiments under various experimental conditions.