An artificial neural network (ANN)-based approach to on-line static security monitoring and assessment of electric power systems

This paper presents an on-line static security assessment scheme that uses artificial neural network (ANN) techniques. The development of this scheme is motivated by the need for a real-time, on-line static security assessment of power systems to aid system operators in maintaining power system security. The purpose of the paper is to serve as an introduction to ANN applications in on-line static power system security assessment. The discussion covers the definition of power system security, a description of regulatory changes that are likely to affect system security, an overview of the static security assessment functions, a background of pattern recognition and ANN approaches to power system security problems, and the presentation of the proposed ANN-based security assessment scheme. Introduction Power system security is defined as the ability of a power system to withstand, without serious consequences, any one of the preselected "credible" disturbances or contingencies [I]. Based on this definition, an operationally "secure" power system is one that has a high probability of maintaining system operation during and after a contingency. Maintaining power system security is accomplished by ensuring that all interconnected generating units are synchronized, and the system state variables (e.g. bus voltages, line flows, frequency, etc.) are within operating constraints. Operating a power system close to its operating constraints degrades the system security and may result in an emergency operating state when a contingency occurs. Power system security has become a critical concern to electric utilities because of the growing power demand. This concern is heightened by the recent regulatory policy changes that are likely to result in increased use of transmission facilities for bulk power transfer. In this challenging environment, power system operators are taking more responsibility in maintaining the security of their respective power systems. To help them carry out their responsibility, they are looking for on-line system security assessment tools that can quickly forewarn them of threatening contingencies. These tools give system operators ample time to select and conduct the required control actions to counter a threatening contingency. An approach to the problem of computational time requirement of on-line system security assessment is by using artificial neural network (ANN) techniques. These techniques have been proposed and demonstrated in small-scale power system models to be a viable tool for on-line security assessment [2,3,18-231. Remlatorv Policv Changes and their Likelv Effect on Electric Power System Security A series of regulatory acts is forcing electric utilities to change the way they fulfill their missions. The earliest of these acts, the Public Utilities Regulatory Policy Act of 1978 (PURPA), introduced the concept of the third party generation via Non-Utility Generators (NUGs) and Independent Power Producers (IPPs). A major concern resulting from this act i$ that many NUGs and IPPs are not under the dire& control of utility control centers. As these nontraditional sources become a larger portion of a utility's generating capacity, the reactive power support capability of the utility is reduced. The Clean Air Act Amendments of 1990 (CAAA) will have a profound impact on the operational policies of utilities, but in a much more direct manner. Utilities may no longer be able to readily implement econon~ic dispatches without regard for the waste emission characteristics of the prime mover fuel. Power plants that do not meet the standards will be forced to undergo costly refurbishment or limit their electric energy production. This condition may result in an emission-constrained dispatch that will reduce the reactive power support of utilities' power systems. The most recent regulatory act is the National Energy Policy Act (NEPA) of 1992. The provisions of NEPA allow Exempt Wholesale Generators (EWGs) to request transmission system access for the purpose of wholesale wheeling. In addition, NEPA grants FERC the statutory authority to order utilities to open their transmission system to third parties. Open access of the transmission system and transmission wheeling will likely cause new power flow patterns, and may significantly reduce the operating Copyright c 19by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved. margins, making the system more susceptible to contingencies. In addition, tie line flow patterns will change dramatically as more wholesale wheeling occurs, and may have a serious impact on transmission line loading. Power System Securitv Concept In evaluating power system security, three modes of security are addressed: (1) steady-state security the condition when the system is operating at a normal steady state condition, (2) transient security the condition of the system during a transient disturbance (measured in cycles), and (3) dynamic security the condition of the system during a disturbance in the order of a few minutes [4]. At present, only the steady-state power system security (i.e. post-contingency state) can be readily assessed by online computer applications. Although such an approach to security evaluation has some limitations, it can be used in estimating the dynamic behavior of a power system by taking static "snapshots" of system states. In 1967, T.E. Dy Liacco proposed a multi-level approach to control problems in maintaining power system security [5]. He designated three emergency security levels or states as preventative (normal), emergency and restorative. Transitions among the operating states are shown in Figure 1. PREVENTATIVE (NORMAL) , 4RESTORATIVE EMERGENCY Figure 1. Dy Liacco's Operating State Strategy