关键词: 绕坝渗流, 异常值识别, DBSCAN

Abstract: Dam bypass seepage is one of the main engineering geological problems in dam safety monitoring, and its safety monitoring results are very important for the stability evaluation of dams. Outlier detection of dam bypass seepage monitoring can reflect the dam stability status from specific dimensions and provide scientific basis for dam management. At present, the outlier detection model used in most dam safety monitoring is based on threshold, which regards the dam bypass seepage as a static problem and ignores the essence of seepage evolution as a dynamic system. To solve this problem, the DBSCAN method is proposed to build the seepage outlier detection model, which realizes the automatic outlier detection by using the tightness of seepage data distribution. Taking the bypass seepage of Lazhai dam as an example, this method is verified, and the results show that this method can accurately identify outliers automatically, which has high application value in practical engineering.

Key words: dam bypass seepage, outlier detection, DBSCAN