Research On Nonlinear Dynamic Model Reduction Of Regional Power Grid

Detailed high-order nonlinear model can describe the practical physical system dynamics well, however, high-dimension and complexity cause a lot of questions in its application, such as computation and analysis, on-line simulation and predictive control, and it even results in some questions can’t complete. Therefore, research about how to decrease the dimension of the regional grid nonlinear dynamic model is significant in academic research and has applied value. This paper employs balanced reduction based on the empirical Gramian method for the nonlinear dynamic model reduction of regiional power grid, and the main research is as follows:Fistly, formulate the nonlinear power system model. Six-order synchronous generator is employed to describes generator’s dynamics, and via the coordination transform between the network, the load and the generator, we get the nonlinear power system model suitable for the model reduction, and the model can be implemented by programing in Matlab.After that, we research the empirical Gramian balanced reduction of the nonlinear dynamic model. Use Karhunen-Loeve decomposition to calculate the principal part of the system, and via Galerkin projection to project the key character to the subspace. To ensure the controllability and observability, employing balanced transformation to transform the original system to its balanced form, and getting the Hankel value of the balanced system, which means the empirical controllable Gramian matrix equals the observable matrix; the Hankel value can be used to determine the dimension of the respective key states, and via Galerkin projection the states responding to the important values can be projected to the subspace to complete model reduction. Based on the Matlab platform, the paper takes the single infinite system as an example to describe and verify the implement process in detail.In addition, to enhance the accuracy of the empirical Gramian balanced reduction, analysis is implemented in effective factors during the formulation of the empirical controllable and observable Gramian matrices, including the excitation control and the governor control from the controllability, and how the states affect the output from the observability, and the four-gen system and the16-gen system are simulated in the Matlab simulation platform.