基于相似测点对比的大坝变形监测数据粗差识别方法

Gross error identification method for dam deformation monitoring data based on similar measurement point comparison

针对混凝土坝变形监测数据粗差检测方法难以区分粗差和环境突变引起的数据跳跃问题,提出了一种基于K-means++距离聚类算法分区,采用OPTICS密度聚类算法结合局部异常因子(LOF)算法识别监测数据中粗差的大坝变形监测数据粗差识别方法。该方法通过K-means++算法实现测点区域划分,联合使用OPTICS算法和LOF算法进行粗差确定,通过对比属于同一分区不同测点的粗差出现时间来判定真实粗差。算例分析结果表明,该方法能有效鉴别变形监测数据中由环境突变引起的数据跳跃,显著提高粗差识别的准确性,降低粗差误判率。

With the gross error detection methods for deformation monitoring data of concrete dams,it is difficult to distinguish between gross errors and sudden data jumps caused by environmental changes.To address this problem,a method for identifying gross errors in dam deformation monitoring data is proposed.This method partitions measuring points using the K-means++clustering algorithm,and employs the OPTICS clustering algorithm combined with the local outlier factor(LOF)algorithm to detect gross errors in the monitoring data.First,the K-means++algorithm is used to partition the measurement point areas.Then,the OPTICS and LOF algorithms are used to determine the gross errors.Finally,the real gross errors are identified by comparing the occurrence time of gross errors at different measurement points in the same area.The case study results demonstrate that the method can effectively identify data jumps caused by environmental changes in the monitoring data,significantly improves the accuracy of gross error identification,and reduces the misjudgment rate of gross errors.