Erika Schäfer

A cascaded control structure for air-path control of diesel engines

In order to comply with increasingly stringent diesel-engine emission legislation, fast and precise control of the turbochargers and exhaust-gas recirculation are necessary. The difficulties lie in certain disadvantageous plant properties such as cross-couplings and non-linearities, the disturbance of the air path by fast operating-point changes and the multitude of air-path configurations that are available. In this paper, a novel model-based approach for air-path control based on cascaded control is proposed. The resulting multi-variable controller has a low sensitivity to the non-linearity of the plant and the disturbances caused by sudden changes in the operating point. Furthermore, the controller is applicable to various types of air-path configuration. This flexibility is demonstrated by a system analysis of the models of two engines which span a broad range of air-path configurations. The performance of the proposed controller is compared experimentally with that of a conventional model-based multi-variable controller, by implementing both on an engine test bench. This comparison confirms that the cascaded controller presented herein can handle the cross-couplings of the system and that it exhibits a lower sensitivity to the non-linearity and the disturbances than the conventional controller does.

Model-Based Injection and EGR Adaptation and its Impact on Transient Emissions and Drivability of a Diesel Engine*

Abstract The emissions of diesel engines are harmful to both humans and the environment. Especially during load steps, the particulate emissions can be excessive. This effect is due to the lack of fresh air caused by the rotational dynamics of the turbocharger. One way to deal with this problem is the model-based limitation of the injected fuel mass based on the estimated cylinder charge. Furthermore, the EGR can be reduced in order to supply more exhaust gas to the turbine and to facilitate a faster acceleration of the turbocharger. In this paper, a method for the combined, model-based reduction of the injection and the EGR is presented. It is based on a dynamic estimation of the cylinder charge. The equations used for the estimator need to be consistent with the ones used for the model-based injection reduction. Otherwise, the injection reduction will not be accurate. An extension to the estimator, which resolves that issue, is presented. The proposed combined injection and EGR limiter is validated with measurements on a test bench. These measurements show how the tuning parameters of the injection and EGR limiter can be used to influence the trade-off between transient emissions and drivability.

Model-Based Online Parameter Optimization

Abstract Changing environmental influences as well as changes in internal parameters of mechatronic systems deteriorate the system performance. A good means to ensure good performance despite these impediments is self-optimizing control. This paper presents a new structure for self-optimizing control systems that forms the basis for combining methods of artificial intelligence with those of “classical” adaptive control theory. In an exemplary simulation on a real-time system, this structure is used to build a self-optimizing pressure-supply system. This realization requires a special optimization algorithm, which constitutes another focus of this paper.