作　　者： (应斐昊）; (邢宁哲）; (纪雨彤）; (纪晨晨）; (李文璟）;

机构地区： 北京邮电大学网络与交换技术国家重点实验室

出　　处： 《北京邮电大学学报》 2017年第S1期108-111,共4页

摘　　要： 针对电力数据网对流量异常检测的时效性要求,提出一种改进的局部异常因子异常检测方法 KTLAD.该方法基于密度进行检测,计算每个流量包与附近流量包的分隔程度,无需预先设置流量的具体异常状态,相对传统方法具有很高的灵活性.仿真结果验证了KTLAD在电力数据网中业务流量异常检测中的可行性,并且有效地降低了时间成本. Due to the efficiency requirements of traffic anomaly detection in electric power data network,an improved anomaly detection algorithm named k-d tree based Lof anomaly detection( KTLAD) based on LOF was proposed. Based on density detection,the algorithm calculated the separating level of each traffic package with nearby ones without pre-set specific abnormal state of traffic. Comparing to the traditional algorithms,the proposed algorithm was more flexible. Simulation results showed that the KTLAD was feasible in traffic anomaly detection in electric power data network and reduced time cost effectively.

关 键 词： 电力数据网 流量异常检测