Diesel engine fuel injection control using a model-guided extremum-seeking method

Abstract

Diesel engine fuel injection control is presented as a feedback based online optimization problem. Extremum seeking (ES) approach is used to address the online optimization formulation. The cost function is synthesized from extensive experimental investigations such that the indicated thermal efficiency of the engine is maximized while minimizing the NOx emissions under external boundary conditions. Knowledge of the physical combustion and emission formation process based on a precalibrated non-linear engine model output is used to determine the ES initial control input to minimize the seeking time. The control is demonstrated on a hardware-in-the-loop engine simulator bench.