摘要
滑坡的阶段划分是滑坡演化过程中的一个重要问题。在本文中,利用DBSCAN算法对滑坡的累积位移序列数据进行聚类,在聚类簇的个数大于阶段数情况下,引入一个统计量M表示滑坡从稳定发育状态变化为不稳定的快速变形发育状态的程度。选择最小的统计值m所对应的滑坡时刻为滑坡演化变形的阶段划分点。实验结果显示,与实际的滑坡阶段分界点吻合,表明DBSCAN算法能有效的对滑坡阶段进行准确的划分。
The division of landslide stages is an important issue in the evolution of landslides. The Densi-ty-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is used to cluster the cumulative displacement sequence data of the landslide in this paper. When the number of clus-tering clusters is greater than the number of stages, a statistic M is introduced to indicate the de-gree of the landslide changing from a stable development state to an unstable rapid deformation development state. The landslide moment corresponding to the smallest statistical value m is se-lected as the stage division point of landslide evolution and deformation. The experimental results show that it agrees with the boundary point of the actual landslide stage. It is shown that the DBSCAN algorithm can effectively divide the landslide stage accurately.
出处
《应用数学进展》
2020年第2期166-171,共6页
Advances in Applied Mathematics
基金
中国地质大学长江流域地质过程及资源环境研究计划(编号:CUGCJ1802).
