Abstract: The information infrastructure and important information systems in the railway industry generate a wide variety of data types, large quantities, and high value density, and different types or levels of railway network data have different levels of security risks. In order to improve the risk assessment mechanism for railway network data, this paper designed a risk level classification method for railway network data based on GMM clustering. The paper extracted key information from the perspectives of data and risk, and constructs a risk information dataset, obtained initial cluster centers through K-means clustering, performed GMM clustering based on mixed distance calculation, and implemented data risk level classification. Through experimental verification, compared with traditional K-means clustering and spectral clustering algorithms, the GMM clustering algorithm has a better clustering effect on railway network data and can more accurately classify the risk level of railway network data, which provide important technical support for further implementing the requirements of railway network data security management.