A Hidden Surveillant Transmission Line Protection Layer for Cyber-Attack Resilience of Power Systems

This article proposes a framework to enhance the resilience of cyber-physical power systems (CPPSs) against cyber-attacks that are capable of bypassing the cyber-based defense mechanisms. To do so, a hidden and local surveillant protection layer is introduced that utilizes isolated measurement devices. Since this surveillance layer relies on local measurements, cyber-attackers cannot affect its performance. However, it requires highly accurate fault detection and classification units (FDCUs) which means requiring additional expenses. Therefore, at the outset, this article employs a deep-learning-based fault detection and classification method using a bidirectional long short-term memory (Bi-LSTM) model to achieve high accuracy with only local transmission line current measurements. The insight and knowledge of the FDCUs are also shared across their neighboring buses through the power-line-carrier communication system. Owing to the need for additional hardware, this system is modeled within a techno-economic framework. The established framework is applied to the CPPS through the evaluation based on distance from average solution (EDAS) method. The EDAS method allows for dynamic adjustments to the integration level of FDCUs based on an analysis of potential cascading failures from various cyber-attack target sets. Extensive simulations conducted on the IEEE 30-bus testbed validate the effectiveness of the proposed framework. The conducted evaluations show that the Bi-LSTM model achieves an impressive accuracy level exceeding 99.66%. This result highlights the robust performance of the proposed surveillant layer and demonstrates its superiority over existing fault detection and classification methods. The scalability of the proposed framework is also confirmed on the IEEE 118-bus testbed.