Abstract: In order to build the distributed general formal description of the high-speed train control system, this paper takes the hybrid modeling method that combines Agent theory with HCPN for the research on the train-to-ground communication scenario in CTCS. The train-to-ground communication scenario is abstracted as MAS. The internal?reasoning process?and the formal definition for the train-to-ground Agents in scenario is given. A?hierarchical structure of HCPN?model is proposed, which visualizes the process of train-to-ground communication, the overall?behavior of MAS and the process of internal reasoning in Agents. The correctness of the model is ensured by analysising the state space report, as well as the feasibility of the model is verified through the multiple simulations on the model which we get how the aperiodic message retransmission interval in the case of different communication failure rates affect the performance of control train in real time during the study scenarios.

Key words: train-to-ground communication, Multi-Agent System（MAS）, agent reasoning, Hierarchical Colored Petri Net（HCPN）

摘要： 为了更好地建立高速列控系统分布式整体形式化描述，采用Agent理论和层次着色Petri网混合建模方法对中国列车控制系统车-地通信场景进行了研究。将车-地通信场景抽象为MAS，给出了场景中车-地主体Agent的内部推理过程和形式化定义，采用CPN Tools工具建立了层次化结构HCPN模型，使车-地通信过程、MAS整体行为以及车-地主体内部推理流程可视化。通过对模型状态空间报告的分析确保了模型的正确性，通过对模型进行多次模拟仿真，分析了该场景中不同通信故障率情况下非周期消息的重发间隔对控车实时性能的影响，验证了模型的可行性。

关键词: 车-地通信, 多智能体系统（MAS）, Agent推理, 层次着色Petri网