Supervised Machine Learning Models for Identifying Critical Grid Situations in Voltage Stability Analyses

IEEE PES Innovative Smart Grid Technologies Conference Europe (ISGT Europe), pp. 1–5

Abstract

As voltage stability becomes more critical, more stability analyses are conducted. This is done by performing dynamic time-domain simulations, which are time-consuming for many grid situations and fault scenarios. This paper examines whether supervised machine learning models can reduce complexity in dynamic voltage stability analyses by identifying critical grid situations based on stability indicators. Dynamic time-domain simulations are performed to pre-classify all grid situations and categorise them into critical and non-critical ones before evaluating the performance of different machine learning models. Overall, all models being assessed achieved high overall performance. Adjusting the misclassification costs further reduces the number of misclassified critical situations. In the future, further investigations on different grid models are needed to evaluate the model’s general functionality.

Authors 3

1. Sophia Tepe Aachen Institute for High Voltage Equipment and Grids

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Institute for High Voltage Equipment and Grids, Digitalization and Energy Economics (IAEW),Aachen,Germany

2. Niklas Brabeck Aachen Institute for High Voltage Equipment and Grids

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Institute for High Voltage Equipment and Grids, Digitalization and Energy Economics (IAEW),Aachen,Germany

3. Albert Moser Aachen Institute for High Voltage Equipment and Grids

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Institute for High Voltage Equipment and Grids, Digitalization and Energy Economics (IAEW),Aachen,Germany