Rapid Transient Fuelling Calibration Method for the PFI SI Engine

This paper presents an experimental calibration method for the feedforward fuelling controller for a PFI SI engine. A recently proposed method [1] is extended from the idle to the torque delivery region and uses a Riccati designed rather than Parameter-Space linear element. Dynamic input signals are applied to air path, load and fuel entering the engine to excite the air-to-fuel ratio dynamics. A nonlinear inverse compensator is obtained directly from the observed input-output behaviour. Least squares black-box identification is used to generate the compensator using an algebraic NARX structure. The resulting inverse compensator not only acts as the feedforward controller but also linearises the fuelling path and therefore makes the system well suited for robust linear feedback control. The feedforward compensator is experimentally demonstrated and subsequently a robust H-infinity feedback controller is designed, implemented and the complete system experimentally validated.

The technique offers several benefits over conventional fuelling calibration. The technique offers a rapid and efficient ‘one-shot’ approach to identify large regions of the conventional fuel map whilst at the same time capturing transient effects. Accordingly, the fuelling calibration can be carried out with significantly less engine testing whilst achieving improvements due to the dynamic nature of the compensator.