Economic Hybrid Model Predictive Control For Optimal Supervisory Control of Parallel HEVs

Abstract

This paper proposes a hybrid model predictive control for optimal supervisory control of a transmission-mounted electric drive (TMED) hybrid electric vehicle using approximate mixed-integer programming. The proposed controller determines an optimal drivetrain mode as well as an optimal torque split of the drivetrain with consideration of future information. The new approach formulates the clutch engagement command and the battery charging/discharging behavior as binary variables and solves mixed-integer quadratic programming using a fast combinatorial integral approximation decomposition technique. The proposed method was validated via a powertrain simulator and showed better performance than the equivalent consumption minimization strategy and the rule-based strategy with decent computation time.