Waschl, H., Alberer, D., and Kerschbaummayr, A., "EGR Reference Allocation for Diesel Engine Air System Control," SAE Technical Paper 2012-01-0892, 2012, doi:10.4271/2012-01-0892.
The control of the engine air system is an essential part for meeting the emission levels of current and upcoming legislation. Up to now different strategies were presented in the literature and also applied on real systems. Starting from simple single-input-single-output structures in combination with feedforward parts leading to advanced multi-input-multi-output approaches. Nevertheless, independent of the used control approach for each of them suitable references are necessary. Although it seems adequate to directly use the emission target quantities in a closed loop air system control, a fast and accurate measurement is seldom available. An alternative is to use intermediate quantities as references, like fresh air mass flow or oxygen concentrations, which represent the state of the air system. However, for control purposes each of these quantities has to be determined, i.e., measured or calculated. Moreover, it has to be considered that each sensor has different dynamics and accuracy in dependency of the given range and furthermore also system dynamics can influence the sensor readouts.In this work a method for controlling the exhaust gas recirculation valve of a diesel engine is proposed, where in contrary to standard approaches not only one but a combination of different reference quantities is used, with the aim to maintain predefined tailpipe nitric oxides emissions. The idea is to use allocation techniques and to combine different measurements with respect to their accuracy and dynamic properties, thus ensuring that in each operating point and time instant the most accurate quantity with respect to NOx is used. The determination of the static accuracy of each quantity was carried out by a NOx sensitivity analysis and the use of Gaussian error propagation. Considering the dynamic properties, response and rise times of the different sensors and physical quantities were taken into account. Finally, the control strategy was implemented and tested on simulation models and on an EU5 passenger car diesel engine on an engine testbed, showing satisfactory results.