With the introduction of the industrial internet, Internet of Things, and big data technology, the interconnection degree of the industrial control cloud network is getting higher and higher, the data interface needs to be gradually standardized, and there are more and more open interface components. Data-based attacks will continue to emerge. The real-time and reliability of access control are essential for trust value updating between network participants. This paper proposes a fine-grained dynamic real-time credibility access control method based on zero trust. Continuous authentication and trust evaluation should be carried out throughout the access control process. The zero-trust evaluation indicators of a rail transit data platform that conforms to the requirements of grade protection 2.0 are established. According to the risk feedback, the current trust level is dynamically updated in real time, and the results are used in the access control model. It can reject unauthorized access, reduce the occurrence of illegal intrusion data leakage and data loss events, and has great value in rail transit data security.