面向电力信息网络的安全态势感知研究

A Study on Electric Power Information Network-Oriented Security Situation Awareness

针对电力信息网络的安全态势精确判断问题,提出一种基于机器学习的安全态势感知方法,并将其应用于实际现场环境。该方法将安全态势感知抽象为分类问题,将实际现场监测设备的记录做为数据源输入到分类器以得到感知结果。基于球向量机设计分类器,并利用量子遗传算法搜索球向量机最优训练参数以提高分类精度。基于KDD Cup 99数据集的实验和系统的实际运行情况表明,该方法在态势感知精度方面优于传统方法。

In allusion to precise judgment of security situation of electric power information network, a machine learning-based method for security situation awareness （SSA） is proposed and applied to actual environment. The proposed method abstracts the SSA as classification problem, and the records of actual field monitoring devices are taken as data source and input into classifier to attain awareness results. Based on ball vector machine a classifier is designed, and using quantum genetic algorithm the optimal training parameters of ball vector machine are searched to improve classification accuracy. Results of experiments based on KDD Cup 99 data set and actual operation of SSA system show that the proposed SSA system is better than traditional methods in the accuracy of situation awareness.