基于DBSCAN的大坝安全监测异常数据检测算法
Abnormal Data Detection Algorithm for Dam Safety Monitoring Based on DBSCAN
李元梦 1李登华 2丁勇1
作者信息
- 1. 南京理工大学理学院,江苏南京 210094
- 2. 南京水利科学研究院,江苏南京 210029;水利部水库大坝安全重点实验室,江苏南京 210029
- 折叠
摘要
为有效识别出安全监测数据中的异常值,考虑环境因素对观测值的影响,提出了一种基于DBSCAN聚类算法的大坝安全监测异常数据检测算法,通过引入数学回归模型获取残差序列,再结合DBSCAN算法对残差序列进行分析,并对大坝安全监测中常见的周期性、趋势性和不规则性数据进行异常检测试验.试验结果表明,该算法对各类异常添加模式下的试验数据查准率、查全率、准确率均达到0.99以上,相比于传统方法具有更好的适用性和鲁棒性.
Abstract
To effectively identify the abnormal values in the safety monitoring,an abnormal data detection algorithm of dam safety monitoring based on DBSCAN clustering algorithm was proposed considering the influence of environmental factors on the observed values.The residual sequence was obtained by introducing the mathematical regression model.And then the residual sequence was analyzed by using the DBSCAN algorithm.The anomaly detection test was carried out on the common periodic,trend and irregular data in dam safety monitoring.The experimental results show that the preci-sion,recall and accuracy of the algorithm are above 0.99 for all kinds of abnormal addition modes,which has better appli-cability and robustness than the traditional methods.
关键词
大坝监测/异常数据/回归模型/DBSCANKey words
dam monitoring/anomaly data/regression model/DBSCAN引用本文复制引用
基金项目
国家重点研发计划(2022YFC3005502)
国家自然科学基金项目(51979174)
国家自然科学基金联合基金项目(U2040221)
出版年
2024
NETL
NSTL
维普
万方数据