Real-time scenario-based stochastic optimal energy management strategy for HEVs

Abstract

This paper proposes a scenario-based stochastic optimal control strategy, considering the stochastic driver behaviors, to deal with energy management issue for parallel HEVs. Firstly, after modelling the dynamic system of parallel HEV including both mechanical and electrical systems, a stochastic model predictive control(SMPC) problem with average constraints is proposed for energy management issue regarding the demaned torque in the prediction horizon as stochastic variable. Moreover, in order to make the proposed problem solvable, two scenarios are chosen and weighted based on the known conditioned transition probability distribustion of demanded torque to transform the original problem into equivalent deterministic nonlinear model predictive control(NMPC) problem. The formulated equivalent problem is solved by employing the Continuation/GMRES algorithm. Afterwards, on-line learning algorithm for updating conditioned transition probabilities of demanded torque is developed since the drive behavior varies as route and environment change. Finally, validation simulation is carried on by a HEV simulator established in the GT-Suite Software.