Imbedding Smart Relays in Large Electric Power Networks: The Scalability Problem and a Possible Solution

As the need to meet the high security and reliability requirements of the electric power grids intensify, it is essential to make the existing relays smarter and more adaptive to changing system conditions. A successful implementation of enhanced relay performance requires both improved logic and the ability to manage computing and communications in near-real time. In this paper, we first summarize our recently introduced hypothesis-testing-based improved logic for over-current relays. The new contribution concerns the computational and communications problem associated with the large combinatorial burden of decision making underlying this new logic when relays are embedded in large electric power networks. We refer to this problem as the scalability problem. In this paper, a possible approach to solving the scalability problem is presented. In particular, this approach leads to: 1) Shorter simulation and training time prior to installing new relays; 2) Less data requirement for training and updating each relay; and 3) Faster dynamic updating with lower communication traffic load. These features are essential for a physical cyber system (PCS) to effectively meet the goals of secure and reliable critical infrastructures, such as a large electric power system.