| In this thesis, a robust approach of Two-Layer Network Equivalent (TLNE) is elaborated. Featuring low-order Marti's line model in surface layer, global searching of low-order deep region networks by Genetic Algorithms (GAs) with passivity and stability constraints and compensation, and Constrained Linearized Least-Square (CLLSQ) optimization with guaranteed convergence as well as the stability and passivity of the obtained model, the robust approach is able to generate the Robust TLNE model with high accuracy and efficiency.; Simulation results and computational time analysis show that the Robust TLNE model is highly efficient in the order reduction of the external system and is suitable for real-time implementation of larger power systems. Two examples and a case study of a realistic large power system---the Alberta Interconnected Electric System (AIES) verify the proposed Robust TLNE approach.; Furthermore, real-time implementation of the second example and AIES is accomplished at 20mus time-step size in a Xeon cluster based real-time simulator in the Real-Time eXperimental LABoratory (RTX-LAB) of the Power Engineering Group at the University of Alberta. A MATLAB/SIMULINK C++ S-function implementing EMTP is programmed to accommodate the Robust TLNE model in the real-time simulator. Real-time simulation results observed from oscilloscope are identical compared to off-line results. |
