Development of a model-based HCCI control strategy for an engine with a fully variable valve train

This paper discusses research activities at the Technische Universität München on the HCCI combustion process, focusing on the development of a model-based control concept with pressure indication. As a first step sensitivity analyses have been carried out to investigate influences of different injection strategies on the combustion and emission characteristics. An optimal injection strategy has been determined and reasonable control variables and ranges corresponding to this strategy were defined. Comprehensive steady-state measurements have been conducted to detect the engine characteristics. In order to limit the experimental effort, principles of DoE (Design of Experiments) have been used to define a methodological approach in the planning of the measurements. Afterwards a multiple-input multiple-output engine model including boundary models for input settings has been designed out of the measurement results. An optimized operation map including constraints for combustion stability and high pressure gradients was derived and used as a basis for a static feed-forward control. Within the overall control concept introduced by this paper the control structure is furthermore enhanced by a closed-loop LQG (Linear-quadratic-Gaussian) output-feedback control based on a linear state-space engine model to track dynamic engine behavior. The control structure has been tested within a nonlinear dynamic engine model and afterwards implemented in the ECU directly at the test bench. This can be a next step towards controlling stable HCCI combustion in ICE's for the future.