Model predictive control of an automotive waste heat recovery system

This paper proposes a model predictive control strategy for an Organic Rankine Cycle based waste heat recovery system. The control strategy uses a prediction model based on gain scheduling of local models, which results in a quadratic program to efficiently calculate the optimal control inputs. To ensure an optimal system operation, the reference values are obtained from a steady-state optimization. To capture a model-plant mismatch, the control concept features an EKF-based estimator of the model uncertainties. Simulations on a validated simulation model show that this control strategy can track the optimal reference very well, even for a large model-plant mismatch.