Discrete adaptive sliding control of automotive powertrains

Automotive powertrain control strategies are a key component of the software design process for vehicle systems. During implementation of control algorithms on real systems, errors often arise that prove costly if they are not detected until the Verification & Validation process. Thus, it is advantageous to mitigate potential uncertainties early on in the design. Specifically, in this paper, we present the control algorithm derivation with the direct incorporation of uncertainty from the system model. We incorporate an unknown parameter representing significant model uncertainty and design to that scenario using a nonlinear, discrete-time sliding control strategy. Additionally, we incorporate an adaptation law to estimate and update the unknown parameter online in order to decrease the control actuation effort. A Simulink representation of the cold start engine emissions process is used as a case study on which the control and adaptation strategy is demonstrated. Simulation results demonstrate that this adaptive formulation yields superior performance over its non-adaptive counterpart by successfully estimating the unknown model parameters and driving tracking error to zero in steady-state.