Stochastic Feedback Combustion Control at High Dilution Limit

Abstract

Cycle-to-cycle variability in the combustion process of spark ignition engines imposes limits when operating at highly diluted conditions. When exhaust gas recirculation (EGR) is used to achieve higher fuel efficiency, the combustion variability (CV) increases due to the reduction in flame propagation speed. A tight control is required to operate close to such limit with minimum variability to maximize EGR benefits. A linear quadratic Gaussian (LQG) controller has been designed to drive the system towards a high efficiency condition with high EGR without deteriorating CV. The controller manipulates spark advance (SA) and EGR-valve opening to target a desired operating condition. The closed-loop system however does not need to be designed only for transient response but also for steady state operation. The directionality of the open-loop system at the target condition is shown to be problematic. Rejection of disturbances along the directions with low plant gain requires large control signals that could drive the system towards misfiring conditions. Moreover, such control commands could be perceived by the driver as torque fluctuations, jeop-ardizing drivability. This limitations at the high dilution limit are discussed and simulated results are provided.