Incorporating Driving Style Recognition Into MPC for Energy Management of Plug-In Hybrid Electric Buses

Vehicle fuel economy is greatly influenced by the driver’s driving style. To achieve remarkable promotion in the fuel economy, a novel predictive energy management strategy (EMS) with strong adaptability to driving styles is presented for plug-in hybrid electric buses (PHEBs) in this article. First, a combined unsupervised and supervised algorithm for the driving style identification is devised based on instant driving conditions. In the algorithm, a multidimensional Gaussian distribution (MGD)-based analysis on factors influencing driver’s driving style provides valuable inputs for local mean
$K$
nearest neighbor (LMKNN) algorithm, and outstanding regression ability of the LMKNN returns exact recognition result and then incorporating the driving style recognition function into the model predictive control (MPC) to formulate a driver-oriented predictive EMS, where estimate distribution and particle swarm optimization (ED-PSO) algorithm is introduced to obtain optimal control sequence over a receding horizon. Finally, simulation validations in the MATLAB/Simulink environment show that the total cost of the proposed strategy is decreased by 12.53% compared to the charge-depleting and charge-sustaining (CD-CS) method, without sacrificing the real-time applicability.