A stochastic logical model-based approximate solution for energy management problem of HEVs

The development of control algorithms to improve the energy efficiency of hybrid electric vehicles (HEVs) has attracted extensive attentions due to the advancement of the powertrain on fuel economy and emission reduction. The objective of the energy management control for HEVs is to distribute the driver demand power between the engine and the motor to maximize the efficiency. Deterministic dynamic programming (DP) has been applied to solve the problem, but it just provides a theoretical optimal solution which cannot guarantee the practical requirements. To deal with the involved time-varying characteristics, significant efforts have been paid to develop model predictive control (MPC)-based online algorithms [1, 2]. However, the nonlinearity in the system causes strong limits to present a general framework to get an exact optimal solution. Moreover, the computational burden is a considerable issue to apply online optimization solutions. The vehicle system explicitly involves the driver behavior which generates stochastic influence to the energy efficiency performance. Recently, extended predictive-based algorithms that integrate the theoretical tools of stochastic system control are investigated [3–5].