基于参数自适应DBSCAN算法的浮标位置数据异常检测-《集美大学学报（自然科学版）》

文章信息/Info

Title:: Buoy Position Data Abnormaly Detection Based on Parameter Adaptive DBSCAN Algorithm

Author(s):: ZHANG Xingliang; XIAO Hong; ZHOU Shibo; Navigation College,Jimei University,Xiamen 361021，China

Keywords:: buoy position; abnormal detection; telemetry and remote control system; DBSCAN algorithm; K-nearest neighbor algorithm; CH index

摘要:: 针对遥测遥控系统采集浮标位置数据时易受外在因素的干扰，提出了一种K近邻优化的参数自适应DBSCAN算法，来检测浮标位置数据中的异常点。通过分析数据集的分布特性生成最优邻域距离值ε和邻域内样本点数量MinPts列表，引入卡林斯基哈拉巴斯指数对列表中的参数进行评分，将最高评分对应的参数作为最优参数，实现DBSCAN算法的自适应聚类。实验结果表明，新算法能够自适应选择最优参数，对浮标遥测位置数据的异常点进行有效检测。

Abstract:: In the process of using the telemetry and remote-control system to collect data,it is easy to be disturbed by external factors and generate abnormal location data.To address this problem,a K-nearest neighbor optimized parameter adaptive DBSCAN algorithm is proposed to detect the anomalies in buoy position data.The algorithm proposed generates a list of optimal distance values ε in adjacent waters and the number of sample points MinPts through the analysis of the distribution characteristics of the dataset,and the introduction of the Calinsky-Harabas index to score the parameters in the list,and the parameter corresponding to the highest score is used as the optimal parameter to realize the adaptive clustering of DBSCAN algorithm.The experimental results show that the proposed algorithm can adaptively select the optimal parameters and realize the detection of abnormal buoy telemetry position data.