Student Research

Power system dynamic security is a growing concern in today's utilities industry. It is generally recognized that current frameworks for dynamic security assessment are not capable of meeting the industry's needs. The interest, therefore, has focused on a new tool of analysis that offers a new framework for assessing power system dynamic security, and which includes the trend in the security status;In this dissertation a new framework for power system security and vulnerability assessment has been developed. Within this framework, system vulnerability is addressed as a new concept for assessing the system dynamic security. The transient energy function (TEF) method was used as a tool to develop this new framework. The new framework can indicate both the present security level using the energy margin [delta]V, and the trend of security status due to the possible variation of a system operating parameter p using the energy margin sensitivity [partial][delta] V/[partial] p. Therefore, this framework can inform us about the weakest point in the system and assess how the changes of the parameter will cause the system to become vulnerable;The indices of vulnerability are determined by establishing the thresholds for acceptable levels of [delta]V and [partial][delta] V/[partial] p; and relating these thresholds to stability limits of critical system parameters;The artificial neural networks (ANNs) technique and the selected multi-layered perceptron architecture are applied to this framework for fast pattern recognition and classification of security status for on-line analysis;The proposed procedure for assessing the system vulnerability and the multi-layered perceptron neural network are tested on the IEEE 50 generator test system. The preliminary results are very promising.

In this research the nonlinear modal interaction and the effect of the interaction on the stressed power system dynamic behavior including excitation control performance are discussed. A systematic scheme based on the normal forms method for the determination of nonlinear interaction between fundamental modes and excitation control modes in a stressed power system is developed;In a stressed power system, the interarea mode phenomenon may occur under large disturbance. Recent investigations revealed that the interarea mode may be among the power system fundamental modes of oscillation associated with the nonlinear modal interaction. If there is significant interaction, the controls will be affected. Because the conventional control system design techniques do not consider the interaction between modes, it is essential to develop a new approach for a clear understanding of the nonlinear modal interaction and its effect on the system dynamic performance;The proposed approach consists of Taylor series expansion, eigen-analysis, normal forms method, and time simulation. In normal form theory, a set of N-dimensional N system modes is said to be resonant of order r (where r is an integer) if [lambda][subscript]j=[sigma][limits][subscript]spk=1N m[subscript]k[lambda][subscript]k and r=[sigma][limits][subscript]spk=1N m[subscript]k for j = 1, 2, ·s, N. In this research work the second-order approxima-tion of the system equations is used. Second-order resonance condition is characterized by [lambda][subscript]k + [lambda][subscript]l = [lambda][subscript]j. If there are no second-order resonances then all the second-order nonlinear terms can be eliminated successively from the vector field using a set of nonlinear state space transformations. The terms of the nonlinear transformation provide important information regarding nonlinear modal interaction;After identifying the modes associated in the interaction and the extent to which they interact, initial conditions for the state variables corresponding to the excitation of the interacting modes are determined using the normal form transformation. These initial conditions are then used to analyze the effect of nonlinear modal interaction on the dynamic system behavior including the excitation control performance;The approach has been applied to two systems which are the four-generator test system and the IEEE 50-generator test system. The results show that excitation control modes interact with low frequency modes and the nonlinear modal interaction can substantially influence the dynamic system behavior.

Approximations to nonlinear system performance are useful in analysis of power-system dynamic response to small or large disturbances. Linear analysis provides a modal picture that describes the system's "natural" response to a disturbance (small or large). Applications, such as modal analysis and energy methods, have taken advantage of linear system approximations. This work investigates the significance of including higher-order terms in the series expansion of the power system's differential equations to the modal behavior of a large, stressed system's transient response. Normal forms of vector fields are used to simplify the system dynamics and to derive an approximate solution to the second-order system in closed form. Interactions of the natural modes of oscillation within the system can be quantified in terms of the solutions for the original-system states. Second-order analysis indicates that many more frequencies of oscillation may have significant influence on the system response. These additional frequencies result from second-order interactions of the linear modes and can not be studied using linear analysis. A methodology based on the normal-form method is developed and utilized to describe the stressed, nonlinear system response by extending linear concepts such as modal dominance and mode-state participation. The relationship between system stress and nonlinearity of the system equations is investigated;The results show that second-order information may be essential to understanding the modal behavior in an interarea-type system separation, whereas linear information may be sufficient for disturbances affecting single plants. Data from the 50-generator IEEE test system is used in this investigation. The contribution of this work is that it includes second-order effects on system performance in a form similar to the linear concept of modal oscillations. In addition, this application of normal forms indicates that higher-order applications may yield additional useful information. Thus, in stressed system conditions where system behavior is not explained using linear analysis, the existing linear methods of control design and placement can be adapted to account for second-order effects. In this manner, the range of usefulness of the existing methods has been extended.

In this dissertation, we apply expert system techniques and Transient Energy Function (TEF) method results for dynamic system security assessment (DSSA). The concept of power system vulnerability combines the effect of contingencies on the security level and its rate of change with the changing system conditions and/or parameters. Vulnerability of a power system can be computed using the results of the sensitivity analysis program of the TEF method. The computation of vulnerability requires a large amount of data obtained from the results of stability analysis programs and this data needs to be organized into a structured knowledge base for DSSA. Further, the power system operators combine the above stability results with their experience and understanding of the power system in deciding the necessary corrective actions. This decision making process, which is based on heuristic rules, can be better solved using expert system techniques than algorithmic techniques;This dissertation describes an implementation of an expert system for DSSA which uses data from the Northern States Power Company (NSP) in Minnesota under certain operating conditions. The NSP system is transient-voltage-limited, and the security index used by NSP is the Twin City Export Margin (TCEM). The TCEM is the difference in MW between the actual export of power from Twin Cities and its allowable limit computed from stability results. This security index is correlated to the normalized energy margin [delta] V of the TEF method. The expert system finds the NSP loading trend and uses this trend to predict the trend of the security index, TCEM. The expert system also adjusts the trend of the controlling parameters so that the security index can be kept at a predefined safe and economic level. The expert system primarily uses a supervised learning scheme and has an interactive user interface with graphic output. The TEF method is used for calculating stability information and the sensitivities of the control parameters on the security index. The knowledge base of the expert system is updated using the results of the TEF method.