Deep Security Analysis Model for Smart Grid

In the face of smart grid attacks generally have a wide variety of problems, we propose a deep security analysis model based on ensemble learning and single classification algorithm to provide a basis for the attack response module. The new model uses the CatBoost, XGBoost and SVDD algorithms to establish a three-level classifier deep security analysis module. We selected the CSE-CIC-IDS 2018 dataset as the validation dataset, and data preprocessing extracted the 16 most important features. The experimental comparison results have achieved good detection results, which can effectively and automatically classify network traffic attacks. The study meets the detection requirements of in-depth security analysis of smart grid systems.