A stacked autoencoder?based convolutional and recurrent deep neural network for detecting cyberattacks in interconnected power control systems

Modern interconnected power grids are a critical target of many kinds cyber-attacks, potentially affecting public safety and introducing significant economic damages. In such scenario, more effective detection early alerting tools needed. This study introduces novel anomaly architecture, empowered by modern machine learning techniques specifically targeted for control systems. It is based on stacked deep neural networks, which have proven to be capable timely identify classify attacks, autonomously eliciting knowledge about them. The proposed architecture leverages automatically extracted spatial temporal dependency relations mine meaningful insights from data coming the systems, that can used as new features classifying attacks. has achieve very high performance when applied real scenarios outperforming state-of-the-art available approaches.