【摘要】 重载运输队列作为现代战争战备物资高效运输方式，有效提升运输能力并降低运输成本。现有队列控制主要关注运动控制特征，忽略了重载特种车辆自身驱动系统构型下系统动力响应特性。基于此，提出了基于多智能体强化学习的重载运输车队队列控制策略，通过控制策略自主式参数优化实现重载队列协同控制，搭建了融合长短时记忆网络的柔性动力需求引导方法，将长期规划策略与短期控制策略解耦，并分别在双层马尔科夫链迭代，建立动力总成元件工况柔性调节控制方法。标准工况试验结果表明：所提出的队列控制策略使队列行驶过程中车头时距保持在1.2 s,动力电池荷电状态维持在35%～65%,并使发动机工作在高效经济区间内，有效提升了重载运输队列的稳定性、耐久性与燃油经济性。更多还原

【Abstract】 As an efficient transportation method for modern war readiness materials, heavy vehicle platoon effectively improves transportation capacity and reduce transportation costs. The existing platoon control mainly focuses on the motion control feature, ignoring the dynamic response characteristics of the powertrain under the configuration of heavy-duty specialized vehicles. Based on this, this paper proposed a platoon control strategy for heavy-duty vehicle platoon based on multi-agent reinforcement learning, where collaborative control of heavy-duty queues was achieved. Through autonomous parameter optimization, a flexible power demand guidance method integrating long short-term memory networks was further constructed. The long-term planning strategy and short-term control strategy were decoupled, and iterated in a double-layer Markov chain, fully releasing the flexible adjustment control of powertrain components under working conditions. The results of the standard driving condition test show that the proposed platoon control strategy maintains a time-headway around 1.2 seconds, maintains the state of charge of battery at 35%～65%, and operates the engine in a high-efficient range, effectively improving the stability, durability, and fuel economy.更多还原