Active Defense Detection Technology for Power System Network Attacks Based on Artificial Intelligence
Huiqin Li, Xu Liyang, Zhao Wenhua, MA Jian-xun · 2023
Traditional methods for detecting network attacks in power systems have a series of problems, such as the need to constantly update the rule base, high false positive rates, and false negative rates. In order to address these issues, the author proposes an active defense detection technology based on artificial intelligence to protect the power system network from attacks. The author proposed 16 new features using log data from smart grids to more accurately describe network activities and abnormal behavior. In addition, the author also proposed an innovative anomaly data processing method and created a model based on the random forest lifting algorithm. The research results indicate that the method proposed by the author performs well in effectively identifying abnormal behaviors such as faults and network attacks in smart grids. This method is expected to solve the problems of traditional detection techniques and improve the security of power systems. The performance of the proposed method is estimated by the accuracy results and it obtains the average value of 0.9018 respectively.