Multi-variable air-path management for a clean diesel engine using model predictive control

Recently, emission regulations have been strict in many countries, and it is very difficult technical issue to reduce emissions of diesel cars. In order to reduce the emissions, various combustion technologies such as Massive EGR, PCCI, Rich combustion, etc. have been researched. The combustion technologies require precise control of the states of in-cylinder gas (air mass flow, EGR rate etc.). However, a conventional controller such as PID controller could not provide sufficient control accuracy of the states of in-cylinder gas because the air-pass system controlled by an EGR valve, a throttle valve, a variable nozzle turbo, etc. is a multi-input, multi-output (MIMO) coupled system. Model predictive control (MPC) is well known as the advanced MIMO control method for industrial process. Generally, the sampling period of industrial process is rather long so there is enough time to carry out the optimization calculation for MPC. However, due to the progress of the computer in the last decade and improvement of the optimization algorithm, the MPC can be applied to 'fast' process such as mechanical systems. The air-pass management system for the diesel engine is assumed a two-input, two-output system in this research. The inputs of the system are the throttle valve and the EGR valve, and the outputs air mass flow and EGR rate. Consequently, MPC was applied to air-pass management system and was modified by adding the disturbance observer to eliminate steady-state error and the compensator for nonlinear characteristics of actuators. The performance of the proposed control system was examined by using an actual testing vehicle. From the experiments, it was shown that an accurate decoupled control of two outputs, i.e. air mass flow and EGR rate, was accomplished by the proposed MPC control.