Hierarchical Data-driven Predictive Control Strategy for Energy Management of Electric Vehicles

For electric vehicles (EVs) equipped with hybrid energy storage systems (HESS), a challenge is reasonably distributing and optimizing load power. To optimize the performances of the vehicle power and battery capacity, This paper introduces an innovative data-driven Energy Management Strategy (EMS) designed for HESS, which combines a hierarchical strategy of fuzzy neural network (FNN) and data-enabled predictive control (DeePC). At the driving level, FNN predicts load power demand by collecting historical speed and acceleration information along with current data. At the energy management level, the EMS adopts a novel data-enabled model, which collects the load power predicted by an upper layer. The DeePC achieves power distribution by controlling the supercapacitor current while maintaining the state of charge (SoC) within the expected range. The simulation results verify the effectiveness of the developed EMS. Compared with long-short term memory (LSTM), artificial neural network (ANN), and support vector regression (SVR), FNN demonstrates the best-fitting performance, while the prediction results are closer to the actual load power. Compared with model predictive control (MPC), the proposed DeePC approximately reduces battery capacity loss by 0.15% and demonstrates well control performance.