Abstract: The time and spatial distribution of railway disconnection and its relationship with precipitation characteristics were analyzed by using the railway disconnection records in Liaoning area and matching the precipitation information in the early stage of disconnection, and the return period of precipitation was calculated. The results show that there are 334 railway outages in Liaoning in the past 11 years, and the annual change trend is consistent with the change trend of summer precipitation in Liaoning. The disconnection was distributed from June to October, and the number of disconnections in August was the highest. The line with the largest number of outages is the Dandong section of the Shendan line, followed by the Fuxin Jinzhou junction section of the Xinyi line and the Jingha high-speed line. The slope of the railway water damage breakpoint is mainly distributed from 0 to 4 degrees. With the increase of the slope, the frequency shows a downward trend. The elevation is mainly distributed in the low elevation area within 300 meters. There is a significant positive correlation between the number of railway outages and precipitation, and the correlation coefficient is as high as 0.86. The main precipitation levels that lead to the disconnection of the railway in summer are rainstorm and heavy rainstorm, and the main precipitation type is local rainstorm. The feature importance of various geographic information elements is sorted, and the top four elements are selected as clustering factors for clustering. The K-means clustering method is used to divide Liaoning into three types of geographical divisions, and the rainfall return period is calculated by using the rainfall in different time periods for the three types of divisions, so as to prevent the railway in summer.