Power System Anomaly Detection Via Ensemble of Encoder and Decoder Networks

Hacking and false data injection from adversaries can threaten power grids' normal operations and cause significant economic loss. Anomaly detection in power grids aims to detect and discriminate anomalies caused by the cyberattack against the power system which is essential for keeping power grids working properly and efficiently. Different types of methods have been applied for anomaly detection such as statistical methods and machine learning-based methods. For machine learning-based methods, we usually need to model the distribution of normal data. In this work, we propose a novel anomaly detection method by modeling the data distribution of normal samples via multiple encoders and decoders. Specifically, the proposed method maps input samples into a latent space and then reconstructs output samples from latent vectors. The extra encoder finally maps reconstructed samples into the latent representations. During the training phase, parameters are optimized by minimizing reconstruction loss and encoding loss. Furthermore, training samples are re-weighted to focus more on missed correlations among features of normal data. Experiment results on network intrusion and power system datasets demonstrate the effectiveness of our proposed method, where our model consistently outperforms all baselines.