Power System Transient Stability Assessment Based on Machine Learning Algorithms and Grid Topology

This work employs machine learning methods to develop and test a technique for dynamic stability analysis of the mathematical model power system. A distinctive feature proposed method is absence priori parameters system model. Thus, adaptability assessment achieved. The selected research topic relates issue changing structure modern systems. key features systems include following: decreased total inertia caused by integration renewable sources energy, stricter requirements emergency control accuracy, highly digitized operation systems, high volumes data that describe operation. Arranging in these new conditions one prominent problems In this study, algorithms based on ensemble (XGBoost Random Forest) were developed low-inertia Transient was analyzed as base function. Features transmission line maintenance used increase accuracy estimation. Algorithms tested using IEEE39. case sample, instability classification XGBoost 91.5%, while Forest 81.6%. increased 10.9% 1.5%, respectively, when topology taken into account.