RBFNN assisted transient instability-based risk assessment of cyber-physical power system

Wide area monitoring and control of modern power system sprawling over large geographical area, has emerged as a viable technology employing phasor measurement units (PMUs) as its backbone. The PMUs are optimally placed to make the whole power system observable. Due to tripping of some generating units, or transmission lines the power balance gets disrupted. As a consequence, the power system becomes prone to risk of transient instability. Also, loss of observability becomes an additional risk due to tripping of transmission lines, which is instrumental for contributing to indirect observability. The purpose of this paper is to develop a predictive approach for identifying vulnerable generating units for proactive control actions to ensure transient stability. The methodology involves use of extended Kalman filter (EKF) for processing real-time PMU data for predicting the dynamic capability of generating unit so as to assess the power swings during transients arising due to contingencies. The radial basis feed-forward neural network (RBFNN) assisted algorithm categorizes generating units into marginal and risky categories based in their trend of swings with respect to dynamic capability curve of generating unit. The case-studies pertaining to IEEE-39 bus system under different contingencies, validate the efficacy of the proposed approach for real world applications. Since, the PMUs communicate synchronized data to PDCs the cyber system is an integral part of the wide area monitoring and control system. Therefore, a cyber physical approach takes into account the loss of observability as a risk from the cyber perspective. Both the risk associated with the physical as well as the cyber layers have been investigated.