• The proposed method integrates raw satellite observation data with RTK to enhance train positioning accuracy by addressing abnormal observations caused by railway environments.
• Zero-baseline testing and dual-frequency GNSS detection enable effective identification of cycle slips and gross errors, improving data quality.
• A data selection strategy for ambiguity fixing increased fixing percentage by 29.2% and reduced standard deviations in East, North, and Up components by 12.7%, 7.4%, and 12.5%, respectively.
• The approach offers a novel perspective for optimizing train satellite-based positioning in challenging railway environments, supporting advanced train control systems.