Real-Time Static Voltage Stability Assessment in Large-Scale Power Systems Based on Maximum-Relevance Minimum-Redundancy Ensemble Approach

Abstract

An integrated approach to static real-time voltage stability assessment in bulk power systems based on mutual information theory is proposed. An advanced maximum-relevance minimum-redundancy (MRMR) algorithm is designed to explore the invisible association between operation variables and the voltage stability margin (VSM). Multiple MRMR procedures with different selected variables are generated in parallel. A set of inter-complementary features is generated one by one using the MRMR criterion that additional VSM information should be reflected in new obtained variables. A functional expression of the relationship between input variables and the VSM is obtained by curve fitting. The performance of the proposed approach is tested on 21- and 1648-bus systems provided by PSS/E. The impacts of training set size, a number of selected feature sets, length of feature sets, and robustness to topology change are studied. Experimental results indicate that compared with other traditional methods, the proposed technique provides faster and more accurate assessment results. Given the proposed method’s efficiency, it is suitable for real-time voltage stability assessment.