Global optimal State-of-Charge reference trajectory prediction and planning under traffic conditions for energy management of plug-in hybrid electric vehicles

The effect of energy management strategy (EMS) of the plug-in hybrid electric vehicles (PHEVs) depends heavily on the state-of-charge (SOC) reference. Meanwhile, the optimality of the SOC trajectory is significantly varied with traffic features of the traffic network. However, they are not effectively used in existing SOC deriving methods. Therefore, a two-level based global optimal SOC planning method is proposed based on traffic conditions. At the upper level, traffic conditions are employed to describe the optimal SOC slope of each road segment over the whole path, and a Bidirectional long-short-term-neural network (Bi-LSTM) is proposed to predict it. Based on this, the global optimal guidance for consuming the electric energy can be obtained. Then, at the lower level, the traffic condition factor and the driving condition factor are constructed to plan the SOC trajectory over each road segment adaptively. In this way, the global optimal SOC trajectory can be derived. Finally, the proposed SOC planning method is incorporated with the adaptive equivalent consumption minimization strategy to track the planned global optimal SOC trajectory to achieve the optimal fuel economy. As shown in the test, the fuel economy has been improved by 7.2% compared to the existing widely used EMS.