An Enhanced PMP-MPC Energy Management Strategy Incorporating Driving Condition Characteristics

The energy management strategy (EMS) serves as a core technology for hybrid electric vehicles (HEVs), directly determining key performance indicators such as drivability and fuel economy. To address the issue of poor real-time performance in model predictive control (MPC) EMS, this paper proposes an improved PMP-MPC strategy based on Pontryagin's Minimum Principle (PMP) to accelerate the single step solution speed within the MPC framework. Firstly, the research conducts driving condition prediction model based on BiLSTM methods to provide predicted vehicle speed sequences within a finite time horizon for EMS. Subsequently, a mathematical model for EMS is established based on PMP, where the shooting method is employed under the MPC framework to iteratively compute finite horizon optimal solution sets satisfying boundary conditions. Finally, A PMP costate initial value prediction model is developed via Backpropagation (BP) methods by fully exploiting future driving conditions enabling rapid determination of costate initial values during MPC rolling horizon optimizations. Results show that the improved PMP-MPC achieves 29.65% reduction in computation time while maintaining equivalent fuel economy performance.