Abstract-In this paper we propose a novel performance index and a multi-layered information structure for monitoring potential quasi-static voltage problems in electrical power systems. The proposed performance index is based on the sensitivity of the smallest singular value of the power flow Jacobian matrix with respect to the corresponding system load level. This performance index is computed at different levels of system aggregation/decomposition, which we define as a "multilayered information structure". This information structure requires little or no information exchange among decomposed subsystems. Moreover, it can be seamlessly implemented in today's horizontally decomposed interconnections. Two multicontrol-area systems are used to illustrate the proposed performance index and information structure.Index Terms-matrix decomposition, power system monitoring, singular value decomposition.
I. INTRODUCTIONHIS paper is motivated by the need for a systematic framework to monitor and control stressed complex power system. As new technical, regulatory and economic drivers emerge in the electric power industry, no such comprehensive framework has been implemented to meet prespecified performance benchmarks over broad ranges of operating conditions. During abnormal conditions, like the August 2003 blackout, problems are two-fold: (a) for individual controller's logic design, there is a lack of adaptiveness over a broad range of operating conditions; and, (b) for system-wide operation, there is a lack of situational awareness of system's deviations outside of normal ranges [1].The objective of this paper is to address the problem (b) above in the context of assessing quasi-static voltage stability.This work was supported in part by the U.S. National Science Foundation ITR Project, and, in part, by the U.S. The problem (a) has been studied elsewhere, for example local feedback design of individual controllers is addressed in [2] and adaptive voltage control schemes in [3]. These papers show that it is generally not sufficient to entirely rely on local voltage control logic, regardless of how advanced the control logic is. In other words, to make system-wide behavior more adaptive, online information is needed to detect when a system is outside of normal operating ranges. Typical information needed about a system is either the minimum singular value of the system Jacobian [3,4], or the direct and quadrature axes components of the generator armature current [2].Both problems (a) and (b) effectively lead to the same key question concerning the minimum information exchange and system decomposition (information structure) for identifying abnormal system conditions, voltage problems, in particular. Implied in this question is the need for a structured approach to defining the information communication which guarantees adequate performance of voltage controllers over a wide range of operating conditions. Because of the communication cost constraints in terms of data gathering and processing, and the administrative boundary...