Attack Detection in Power Distribution Systems Using a Cyber-Physical Real-Time Reference Model

This paper develops a novel intrusion detection system for power distribution systems that utilizes a cyber-physical real-time reference model to accurately replicate the complex behavior of power distribution components for attack detection. The proposed intrusion detection system analyzes the consistency of the physical data from sensor readings and control commands in contrast with their digital counterpart obtained from the real-time reference model. Therefore, a residual-based online attack detection mechanism that utilizes the chi-square test is able to determine the presence of malicious data. The proposed mechanism is implemented on a hardware-in-the-loop simulation testbed on a test power distribution system with distributed energy resources and energy storage systems. The results show that the proposed solution is able to quickly detect multiple types of cyber attacks for different levels of accuracy of the real-time reference model. The simulations illustrate that the proposed approach outperforms conventional attack detection mechanisms such as one-class classifiers and residual-based approaches that utilize Kalman filters or neural networks.