Model-Predictive Control of Diesel Engine Airpath

With the new fuel economy and emissions requirements, multiple air system actuators must be used to meet transient power demands and change the engine operating point while minimizing emissions. Due to coupling in system dynamics, traditional single-loop PI controls prove to be insufficient. Also, multiple-input multiple-output (MIMO) control techniques are increasingly seen as an attractive alternative. Model predictive control offers a structured and intuitive way to accomplish MIMO design. This presentation shows how to design model predictive controllers for simultaneous control of boost pressure and exhaust gas recirculation mass flow targets in the presence of driver fuel demand and engine speed changes using VGT and EGR. A high-fidelity engine model is linearized at multiple operating conditions using system identification methods. The resulting linear models are used for the design of a gain-scheduled model predictive controller. This controller is then validated in nonlinear simulation.