Using PCA to Design an Anomaly Monitoring System for Power Grid Cryptographic Equipment

The network equipment and operation devices in the smart grid generate a large amount of communication interaction data and related system logs every moment. These logs reflect the working status of the relevant equipment and devices in the grid system, and most of them are working fine. Some equipment and devices may fail to work, while others may perform irregularly. It is wise to find out the corresponding irregular logs efficiently in the massive normal equipment logs of the power grid system and extract the addresses of abnormal equipment for statistics display and alarm alerting. We have designed and experimented to find that a simple tool like PCA (Principal Component Analysis) can accomplish that purpose of anomaly detection. Hoping to improve the efficiency of grid network operators, we propose an anomaly detection scheme that uses the PCA algorithm to find outlier data by preprocessing logs from cryptographic equipment in the smart grid network and then comparing the outlier data with the original data to extract the IP addresses of the faulty equipment.