Abstract: AAiming at the problem that the data of railway station signal layout plan cannot be reused due to different data formats, this paper presented a data mining based method for extracting information from railway station signal layout plan. The paper constructs the model and code of railway signal engineering legend, and clusters the graphic data into graphic groups based on RV-DBSCAN algorithm, constructed a model through C4.5 decision tree to identify railway signal legend. The test results shows that the FMI score of the clustering method is 0.9860, and the accuracy rate of the classification algorithm is 95.64%. It can accurately identify the legend symbol data in the layout plan, and provides a general data interface for the secondary use of the layout plan information.