采用Boosting方法预测电力信息网络的威胁态势  

A Boosting Algorithm Based Method to Predict Cyber-Threats Situation of Power Information Network

作　　者：徐茹枝[1] 王婧[1] 朱少敏[2] 许瑞辉[1] 

机构地区：[1]华北电力大学控制与计算机工程学院,北京市昌平区102206 [2]北京市电力公司,北京市朝阳区100020

出　　处：《电网技术》2013年第10期2825-2829,共5页Power System Technology

摘　　要：威胁态势预测可以有效反映电力信息网络在未来时刻的宏观安全状况。为实现威胁态势的精确预测,提出一种基于AdaBoosting方法的网络威胁态势预测方法。该方法采用威胁态势值描述电力信息网络的宏观安全态势,并将威胁态势值的预测抽象为回归问题,进而利用AdaBoosting方法求解。该方法先利用滑动时间窗口将威胁态势值构造成时间序列样本集,再将样本集输入到AdaBoosting方法中训练,以得到回归分析模型,并利用该模型完成威胁态势预测。最后基于现场数据的验证性实验证明了所提方法的有效性。The prediction of cyber-threats situation can effectively reflect the macroscopic security situation of power information network in the future time. To realize the accurate prediction of cyber-threats situation, an AdaBoosting algorithm based cyber-threats situation prediction method for information network is proposed. In the proposed method, the values of cyber-threats situation are used to describe the macroscopic security situation of power information network, and the prediction of macroscopic security situation is abstracted to a regression problem, and then the regression problem is solved by AdaBoosting algorithm. Firstly, using the sliding time window a time series sample set is constructed by cyber-threats situation values; then the sample set is input into AdaBoosting algorithm to be trained to obtain a regression analysis model; finally the prediction of cyber-threats situation is completed by the regression analysis model. Finally, the effectiveness of the proposed method is verified by results of replication experiments based on field data.

关 键 词：电力信息网络 网络威胁态势 预测 AdaBoosting方法 支持向量回归 回归问题 滑动时间窗口 

分 类 号：TM734[电气工程—电力系统及自动化]