Study On Emergency Control Of Power Systems Based On On-Line Dynamic Equivalence And Energy Function Method

The time-domain simulation based on numerical integration technique is the most commonly used tool to investigate transient stability in the plan and operation practice of modern bulk interconnected power systems. But the simulation method is disadvantageous because it is time-consuming and unable to provide any quantitative index to measure system stability. So it is necessary to develop a more effective tool to assess transient stability of power systems, which permits rapid stability analysis with a certain precision and makes possible transient stability analysis and control in real-time. On account of the characteristics of the complicated power systems, a two-stage scheme is proposed in this dissertation to deal with power system transient stability. The dynamic equivalencing and energy function method based on Lyapunov's stability theory is combined to develop a tool for assessing power system transient stability with on-line application value. In Chapter One of this dissertation, the historical and current status of the study on the above aspects is summarized, and a new technical route is proposed with combination to practical requirements. Some innovation points throughout the dissertation are briefly listed.A salient feature of bulk interconnected power system lies in its large scale, and as a result it poses great burden on even off-line transient stability analysis for detailed system model. It is the case mainly because the implicit integration method is widely used to ensure precision in simulation, and the solution of high-order differential and algebraic equation sets involved is memory- and time-consuming. To alleviate the computation load, dynamic equivalencing method is adopted to reduce the scale of interconnected system, which means reducing the model order of nonlinear system to be simulated. Two dynamical equivalencing methods, coherency-based equivalents and modal equivalents, are developed in history. In coherency-based equivalents, the simplified power system model is approximately simulated by numerical method, and coherency is recognized according to generator rotor angles thus obtained. Coherent generator groups away from fault site are equivalenced respectively, and the whole network is reduced to its equivalent form. In modal equivalents, power system model is linearized around a steady-state operation point,and modal analysis is made on the linearized system. Then transient stability is analyzed based on the complete system model in mode-decomposition form or in reduced-mode form. In the two methods, whether the approximate numerical simulation for simplified model or the modal analysis for the linearized system is time-consuming, which makesthem only applicable to off-line stability analysis. Because regionalization of dynamic behavior is often observed in the simulation of power system subjected to faults, a region-oriented dynamic equivalencing method for interconnected system is devised in the dissertation on the assumption of regional coherence. In the first place, the branch parameters in the equivalent subnetwork of the external subsystem are determined from the pre-fault steady-state operation point on the basis of linear circuit theory. In the second place, the coherent generators in external subsystem are replaced by the equivalent one, whose parameters are obtained through system identification process. Parameter identification process goes as follows. A fifth-order model is adopted for the equivalent generator, the measurement of the current injection into the boundary buses of external subsystem is taken as the Input, and the boundary buses voltages of external subsystem are taken as the Output. Therefore a system identification model including Input and Output can be built. If the Input is given, the sum of squares of the differences between the measurement values and computation values in equivalent model is taken as objective function, a nonlinear least-square optimization model can be constructed. Because some comparative quantity relations between some para...