With the continuous advancement of the construction and development of urban rail transit networks, people's demand for transportation capacity is increasing day by day. Therefore, the shortage of on-board computing power has become one of the important issues, making it necessary to explore new solutions. In recent years, Edge Intelligence(EI) has emerged as a new field. Through edge intelligence, complex computing tasks can be offloaded to trackside computing servers, fundamentally liberating the computing power of onboard equipment. In this case, we can focus the onboard devices on performing simple, low energy computing tasks, while leaving most of the computing work to edge servers. Based on the above idea, this paper proposes a new train autonomous control system, which uses Google Kubernetes high reliability edge computing platform to realize the train autonomous control algorithm. In addition, we use the Linear Quadratic Gaussian(LQG) algorithm to model the train autonomous control process and utilize cloud security computing to ensure the high reliability of the entire system. At the same time, due to its ability to effectively avoid the impact of local faults, the system also exhibits excellent performance in terms of communication packet interval delay performance.After extensive experimental verification, we can conclude that the proposed train autonomous control system has high operational reliability and data security, as well as low communication packet delay performance. This result further proves that using the autonomous train operation control system can significantly improve the efficiency and safety of train operation, thereby improving the quality of train operation.