Extreme-environment-aware adaptive energy management strategy for heavy-duty series hybrid electric vehicles based on data-driven method

The energy management strategy (EMS) plays a crucial role in series hybrid electric vehicles (SHEVs) by coordinating power distribution, ensuring stable and efficient vehicle operation. Heavy-duty SHEVs operate in extreme environments during long-distance plateau transportation, accompanied by atmospheric pressure changes. These changes can lead to substantial changes in engine characteristics, which significantly weaken vehicle operational stability and fuel economy. Therefore, developing an adaptive EMS that can manage changes in power and fuel consumption characteristics of the engine in extreme environments is pressing. In this study, an extreme-environment-aware adaptive EMS for heavy-duty SHEVs based on data-driven method is proposed. First, backpropagation neural networks model the nonlinear characteristics of the engine, generating a real-time model that accurately captures power and fuel consumption characteristics of engine. Then, dynamic constraints update mechanism is established to real-time reconstruct the operational stability envelope based on engine states and optimal control sequence during iterations. Next, a data-driven dynamic cost function in EMS is designed using statistical regression estimation from historical data, achieving adaptive optimization via real-time coefficient adjustments based on the environment. Additionally, the nonlinear multi-objective optimization problem is convexified and solved through the damped alternating direction method of multipliers, which incorporates damping to enhance convergence stability and efficiency. Finally, through verification, the engine speed fluctuation is limited to 2 %. The fuel economy is enhanced by up to 4.98 %, while the average computation time is decreased to 1.74 ms. Hardware-in-loop tests showed max errors of 2.28 km/h for speed and 7.96 kW for power, ensuring real-time performance and optimality.