Steady-State Data-Driven Dynamic Stability Assessment in the Energy Management System

Abstract

The extensive research on dynamic security assessment stability prediction has focused on data preprocessing techniques to improve accuracy because it was assumed that high-resolution postfault data exist. For practical users, the acquisition and application of high-resolution measurement data present significant challenges. Installing phasor measurement units on all power system nodes is deemed impractical due to high costs. In this work, we aimed to develop a rotor angle stability prediction model using steady-state data that can be easily generated from current energy management system. Note that the steady-state measurement data refer to a pre-contingency operation condition characterized by real and reactive loads, generation levels, flows, as well as voltages and angles. The proposed framework comprises three stages: it finds physical meaning from the extended equal-area criterion to move away from the black-box approach, proposes a feature data extraction strategy to reduce the dimensionality of the input space in the support vector machine, and partition time-series power flow data by month to consider system topology changes. By utilizing 5-min-interval power flow data, unstable cases are determined, and two main feature data are extracted to train the support vector machine. The proposed model can be implemented in the energy management system for industrial use, and the obtained results showed the effectiveness of the proposed framework in responding to a critical line fault event in real time