Anomaly Detection and Processing in Artificial Intelligence for IT Operations of Power System
Yiyu Xia, Jixiang Lu, Yun Li, Bin Zhang, Hao Li, Feng Xie, Shaobo Liu, Chunlei Xu · 2019
In recent years anomaly detection has been wildly applied in many fields, from zero day attack detection to insider threat detection, from situational awareness to intrusion detection. For power system, secure and stable operation is indispensable and it takes electric utility staff huge amount of time. Naturally, artificial intelligence for IT operations (AIOps) especially anomaly detection can also be used to find out unusual behavior discord with expected pattern in this field. In this paper, we propose an intelligent system that first conducts a joint time series detection to identify outliers or anomalies on the basis of statistical judgment and machine learning, and then automatically discovers those anomalous functions in the method of statistical analysis. The result indicates that the implementation of our system is able to largely reduce labor costs, improve automation and efficiency of power system operations and maintenance.