Discrete sliding controller design with robustness to implementation imprecisions via online uncertainty prediction

Analog-to-digital conversion (ADC) is one of the main sources of controller implementation imprecisions due to sampling and quantization. In this paper, a new control approach is developed to mitigate the ADC imprecisions by (i) identifying the ADC imprecisions in the early stages of a controller design cycle, (ii) developing a mechanism for real-time prediction of uncertainties due to ADC, and (iii) incorporating the predicted uncertainties into the controller design. To this end, a generic online technique is developed to predict sampling and quantization uncertainties on measured signals. Then a discrete sliding mode controller (DSMC) is designed for incorporation of ADC uncertainties. Inclusion of predicted imprecisions into the DSMC requires the knowledge of propagated uncertainties on control signals. An experimental approach is proposed for a general class of nonlinear systems to estimate the propagated ADC uncertainties on control inputs. The proposed control approach is illustrated on an automotive engine control problem. The designed controller is tested in real-time in a processor-in-the-loop (PIL) setup using an actual electronic control unit (ECU). The results show that the proposed control approach significantly improves the robustness of the controllers to ADC imprecisions. This provides up to 90% improvement in the performance of the controller under ADC imprecisions compared to a baseline sliding controller.