Online mixed-integer optimal energy management strategy for connected hybrid electric vehicles

Abstract

In this paper, an online mixed-integer optimal energy management strategy is proposed for connected hybrid electric vehicles. Firstly, a predictive framework is constructed based on the backpropagation neural network, aiming to predict the future information utilizing the connected vehicle technology. Subsequently, for the mixed-integer programming problem in the predictive horizon, a novel optimal algorithm is proposed in the predictive framework. Finally, the proposed strategy is verified under both simulation and hardware-in-the-loop system environments. The results show that the proposed strategy reduces fuel consumption by 25.34% and 1.13% compared with the rule-based EMS and equivalent consumption minimization strategy (ECMS)-based EMS, and reduces fuel consumption by 25.79% and 1.78% compared with the rule-based EMS and ECMS-based EMS in two typical conditions. The proposed strategy can reduce 84% computation time than the particle swarm optimization-based EMS in the same typical condition. Using real-word conditions, the proposed strategy can reduce fuel consumption by 9.2% compared with ECMS-based EMS. The proposed strategy achieved satisfactory results in a hard-in-loop experiment.