Emissions of modern combustion engines are decisively dependent on the precision of mass flow control within the engine's gas path. The mass flows are regulated by gas path actuators. A variety of non-linear and time-varying effects have an impact on the dynamic behaviour of gas path actuators. Conventional linear control approaches, which are gain-scheduled by parameter maps, cannot systematically take into account these influences. This results in limited accuracy and performance of the single actuator control loop and, thus, of the super-ordinate gas mass flow controller. Additionally, due to strong dependency on operational conditions, high calibration and test efforts result during the development process. To over-come these problems, a methodology for modelling and identification of gas path actuators and the disturbing forces acting on them is presented in this paper. The resulting models are integrated into a nonlinear adaptive control and disturbance compensation scheme. For this purpose, the Exact Linearization (EL) is combined with joint state and a parameter estimation by an Extended KALMAN-Filter (EKF). This eliminates the main disadvantage of EL, the lack of robustness against model uncertainties. Systematic compensation of the disturbing influences and non-linear time-variant effects leads to widely linear time-invariant dynamics of the resulting control loop combined with low disturbance sensitivity. The concept is tested on a real-time rapid prototyping hardware and evaluated by an experimental comparison with a flatness-based two-degree-of-freedom control structure and the classical gain-scheduling PID series controller. Climatic chamber tests including worst-case scenarios as well as near-series driving experiments prove the performance and robustness of the suggested approach.
展开▼
