| the advanced metering system adds information system into the power system, it uses smart meters, two-way communication system, the user household network and metering data management system to establish communication links with the users, which can provide dynamic billing functionality. Measurement and visibility capabilities provided by the advanced metering system can significantly enhance the existing electric power company operation mechanism and the asset management processes, therefore, the advanced metering system has important development trend and value.The security requirements of the advanced metering infrastructure arediscussed in this paper. The attack tree of advanced metering system isanalyzed, based on the study of the intrusion detection systems for advanced metering infrastructure. The electrical abnormalities in intrusion detection technology based onadvanced metering infrastructure are studied. The advantages and disadvantages of traditional electricity anomaly detection and anomaly detection based on machine learning electricity are summarized. By means ofanalyzing the load models of electricity consumption in different users, the differences of user consumption models in differenttypes of time are analyzed. In this paper, the detection scheme based on One-Class Support Vector Machine is proposed, and the simulation experiments based on the program is conducted. The measured data in the platform is obtained though the smart metering system in Yunnan Power Technology Park’s an experimental platform. Meanwhile,the detection rate for the scheme iscounted. The results showed that the higher classification accuracy can be obtained in case of small sample size. The data required is collected from the metering data management system, without theimprovementof smart meters, two-way communications systems and user household network, it can take full advantage of the hardware of the advanced metering system. Program required less training time, without classifying anomaliesdata in manual, which can effectively reduce the cost of applications. |
