Tree-Based Approaches for Parameter and Load Margin Estimation of Power Transmission Systems

Real-time voltage stability assessment has become a prominent area of study due to the increase in load demand and the diversity of non-linear loads. Additionally, the large amount of data available allows the use of data-centric solutions. In this context, this paper proposes a comparative analysis of tree-based methods for Load Margin Estimation. The presented methodology uses a reduced number of features, avoiding the use of power flow measurements, and relies on PMU measurements from only a few buses in the system. The method was also tested considering the presence of random and systematic errors in phasor measurements. The proposed method is evaluated on the IEEE 14-bus test system. Results show that the Load P Margin (LPM) can be predicted with an error of less than 1% in the noiseless case and less than 2% in the presence of noise.