Non-linear model predictive control of a power-split hybrid electric vehicle influence of conclusion of powertrain dynamics

Hoseinali Borhan, Ardalan Vahidi, Wei Liang, Anthony Phillips, Stefano Di Cairano, Ming Kuang, Ryan McGee

CHALLENGE - In hybrid electric vehicles (HEV) optimal power management is a challenging task. In this paper a nonlinear predictive model (MPC) is presented to minimize the energy consumption during the trip considering equivalent fuel cost function for battery energy usage and fuel consumption plus the equivalent cost of battery. The MPC requires two different calibration parameters which must be tuned. 

SOLUTION - The MPC implemented in Matlab has been included in a modeFRONTIER workflow to calibrate two tuning parameters. One is used when the engine is on and one when the engine is off, enabling to consider the start and stop functionality in the predictive model. The F-Simplex algorithm has been used to calibrate the curves to obtain the best fuel economy.

BENEFITS -   Using modeFRONTIER enables the automatic calibration of parameters of the MPC saving a considerable time running the simulations on the cluster and exploiting the combination of real and virtual optimization capabilities of the F-SIMPLEX algorithm.