Hierarchical optimization control strategy of hybrid electric vehicle platoon under communication failure

This research proposes a DRL-based hierarchical optimization control strategy for connected HEV platoons through a cloud platform, addressing strong nonlinearity and communication failure issues. The strategy uses a signal-interference-plus-noise ratio model to detect network failures by considering distance, signal path loss, and wireless interference. The high-level control employs distributed model predictive control (DMPC) to generate desired commands for platoon driving during network failures. Meanwhile, the low-level control leverages prior knowledge of the engine's optimal brake fuel consumption curve and battery characteristics to optimize energy management through knowledge and data fusion. To enhance energy planning efficiency, a PER-D2PG intelligent algorithm is introduced, integrating priority experience replay and dueling networks into DDPG. A trusted Markov decision process and a self-learning energy optimization framework are also established. Numerical results demonstrate that the proposed strategy effectively adjusts engine and motor power distribution, achieving vehicle car-following, safety, and energy-saving goals.