基于FARIMA模型的电网数据服务器流量异常识别  

Recognition of abnormal flow of power grid data server based on FARIMA model

作　　者：王翀 周雨迪 蔡新雷 WANG Chong;ZHOU Yudi;CAI Xinlei(China Southern Power Grid Digital Grid Research Institute Co.,Ltd.,Guangzhou 510600,China;Electric Dispatch and Control Center,Guangdong Power Co.,Ltd.,Guangzhou 510610,China)

机构地区：[1]南方电网数字电网研究院有限公司,广东广州510600 [2]广东电网有限责任公司电力调度控制中心,广东广州510610

出　　处：《电子设计工程》2023年第12期115-118,123,共5页Electronic Design Engineering

摘　　要：当前电网数据服务器流量异常识别方法的平均时延过长,丢包率过高,因此提出了基于FARIMA模型的电网数据服务器流量异常识别方法。利用多重分形测度方法提取一个维度的电网数据流量特征,对存在较大方差的流量特征进行描述,利用FARIMA模型的回归算法计算电网数据服务器流量阈值,以此分析电网在正常运行条件下的数据服务器流量行为,并对此时的流量阈值进行检测,利用FARIMA模型与参数估计方法实现电网数据服务器流量异常识别。实验结果表明,所提方法能够有效缩短平均时延,降低丢包率。At present,the average delay of the traffic anomaly identification method of power grid data server is too long and the packet loss rate is too high.Therefore,a traffic anomaly identification method of power grid data server based on FARIMA model is proposed.The multifractal measure method is used to extract one dimension of power grid data flow characteristics,describe the flow characteristics with large variance,and calculate the power grid data server flow threshold by using the regression algorithm of FARIMA model,so as to analyze the data server flow behavior of power grid under normal operation conditions,and detect the flow threshold at this time.The FARIMA model and parameter estimation method are used to identify the abnormal flow of power grid data server.Experimental results show that the proposed method can effectively shorten the average delay and reduce the packet loss rate.

关 键 词：FARIMA模型 电网数据 流量异常 异常识别 

分 类 号：TN99[电子电信—信号与信息处理]