On-line fuel-efficiency optimization of Diesel engines using constrained multivariable extremum-seeking

Advanced control concepts are required to maximize the fuel-efficiency of Diesel engines and to comply with emission legislation. The aim of this paper is to develop a controller that fully utilizes the fuel-path and air-path hardware, and is adaptive for real-world disturbances. The control design comprises a feedback control system, which uses air-path actuators and fuel injection settings to robustly track engineout NO_x emission and engine power demands, as well as two parameters that are known to influence the NO_x emission versus fuel consumption trade-off; combustion phasing and pumping losses. Furthermore, an on-line implementable convex cost criterion is proposed that evaluates the injector opening time to obtain a cost output which is equivalent to fuel efficiency. A multivariable constrained extremum-seeking method is applied to minimize the cost output, as a function of combustion phasing and pumping losses, subject to constraints on the tracking performance of the low-level control system. The control design is implemented on a Euro-VI Diesel engine equipped with exhaust gas recirculation and a variable geometry turbine, as well as in-cylinder pressure sensors. The experiments demonstrate the effectiveness of the applied extremum-seeking approach in finding the (constrained) optimum, for a constant engine speed and torque.