Combined pole assignment and mean value engine model multivariable decoupling control

Abstract

This article considers a combined pole assignment and multivariable decoupling control algorithm using discrete-time, non-minimum state space (NMSS) methods. In contrast to earlier research based on low-order linear models, the approach is applied to a nonlinear mean value internal combustion engine model with three control inputs, namely the throttle plate angle, injected fuel mass flow and spark advance angle. The controlled outputs are the air mass flow pressure, crank shaft speed and air–fuel ratio (AFR). It is well known that, for example, regulating the AFR to the stoichiometric value (i.e. $14.7$ ) leads to a desirable balance between power output and fuel consumption, while reducing pollutant emissions. In this regard, the linear NMSS approach is straightforward to design for a range of performance requirements and yields comparable results to a more complex benchmark sliding mode control system. Furthermore, it retains a similar implementation structure to current production units, which are typically based on conventional proportional-integral compensation. The robustness to changing operating levels and disturbances, including an air leakage signal, are evaluated in simulation.