分布式光纤预警系统同质序列数据异常模式挖掘方法  

Method for mining abnormal pattern of homogeneous sequence data in distributed optical fiber early warning system

作　　者：赵海燕[1] 杜丽娟[1] 刘琨[1] 刘建国[1] ZHAO Haiyan;DU Lijuan;LIU Kun;LIU Jianguo(College of Applied Science&Technology,Beijing Union University,Beijing 100012,China)

机构地区：[1]北京联合大学,北京100012

出　　处：《激光杂志》2022年第9期134-138,共5页Laser Journal

基　　金：北京市自然科学基金项目(No.7043082);北京联合大学校级基金资助(No.YB2021113)。

摘　　要：采用目前方法挖掘分布式光纤预警系统中同质序列数据时,存在挖掘失败率高的问题,为此,设计一种分布式光纤预警系统同质序列数据异常模式挖掘方法。采用经验模态分解算法提取分布式光纤预警系统数据的特征,降维处理同质序列数据,并通过请求循环平均异常度、浏览时间平均异常度和序列比对平均异常度这三个角度,检测同质序列数据中存在异常模式,进行最大频繁序列模式挖掘。实验结果表明,所提方法的挖掘失败率低,错误率低,正确率高,预警精度高。At present,there is a problem of high failure rate in mining homogeneous sequence data in distributed optical fiber early warning system.Therefore,a method for mining abnormal patterns of homogeneous sequence data in distributed optical fiber early warning system is proposed.The empirical mode decomposition algorithm is used to extract the features of distributed optical fiber early warning system data,and the homogeneous sequence data is reduced in dimension.The abnormal patterns in homogeneous sequence data are detected from three angles:average anomaly degree of request cycle,average anomaly degree of browsing time and average anomaly degree of sequence alignment,and the maximum frequent sequence patterns are mined.Experimental results show that the proposed method has low mining failure rate,low error rate,high accuracy rate and high early warning accuracy.

关 键 词：分布式光纤 预警系统 同质序列 经验模态分解算法 数据降维 异常模式挖掘 

分 类 号：TN929[电子电信—通信与信息系统]