Incorporation of implementation imprecision in automotive control design

Implementation of control software on physical systems requires the use of hardware that invariably introduces imprecisions. In particular, the use of an analog-to-digital converter (ADC) introduces uncertainties in measured signals due to sampling and quantization effects. These imprecisions can have a significant effect on the tracking performance of the controller, to the point that it no longer meets the desired targets. This leads to costly design and verification iterations. In this paper, a generic methodology for modeling the uncertainty introduced by sampling and quantization is developed. The resulting values are then propagated through the system's state equations to determine the overall uncertainty bounds on the system model. This methodology is illustrated on the cold-start emission control problem. Verification is performed to ensure that the bounds have been accurately estimated. Finally, a sliding mode controller developed for the cold-start phase of an engine is modified to incorporate these bounds. Simulation on a real engine control unit (ECU) verifies the modified controller is more robust to implementation imprecision.