Dynamic matrix control applied to emission control of a diesel engine

Abstract

In this paper we suggest a hierarchical control scheme, applicable to engine control. The chosen setup facilitates the simultaneous control of emissions and torque. On the top level a standard PID controller is installed, setting the injection quantity in order to reach the demand torque, which is a prerequisite to follow a given load profile. The second level is dedicated to the emission control. We apply dynamic matrix control (DMC), which is a specific form of model predictive control. DMC stands out by a very simple way of modeling the controlled process which is represented by step responses. Constraints on the absolute values and maximum rates of change are applied to the manipulable system inputs to cope with given hardware limitations. Moreover, we incorporate so-called operational constraints, thus constraining the input variables to certain polytopic operating regions. Thereby areas with high HC emissions can be excluded in advance. Furthermore we include the predefined demand torque as input to the DMC. With the known impact of transient torque changes on the emissions, the DMC can act in an optimal way with respect to the given load profile. To cope with the nonlinearities of a combustion engine, we apply a network of several DMCs, scheduled by engine speed and load as well as by the manipulable actuators. We also propose a procedure to reduce the amount of modeling data, by introducing a parameterized formulation of the measured step responses. The proposed control concept is then evaluated on a realistic, physically-based engine model for several representative driving sequences, amongst them the well-known New European Driving Cycle (NEDC) and the highly dynamic Worldwide Harmonized Light Duty Test Cycle (WLTC). The advantages of DMC against a conventional PID control are finally summarized and demonstrated on a typical load step.