A Cascaded Economic Model Predictive Control Strategy for a Diesel Engine using a Non-Uniform Prediction Horizon Discretization

This paper presents an economic model predictive control (eMPC) controller for the diesel engine airpath. The control objective is to maximize fuel economy while maintaining drivability and reducing emissions by manipulating the exhaust gas recirculation (EGR) valve, throttle, variable geometry turbine (VGT), and the fueling rate. This is achieved using a cascaded architecture wherein an upper level economic MPC controller controls the fueling rate and generates EGR rate and intake manifold pressure setpoints for a lower level nonlinear MPC airpath controller which manipulates the actuators. A non-uniform prediction horizon discretization is used to manage computational complexity while capturing dynamics with multiple timescales. Simulation results demonstrating emission constraint satisfaction and stable operation over drive cycles are presented.