Skip to content

Research1 min read

Machine learning methods classify power system contingencies using data preprocessing

This study applies machine learning algorithms to classify power system security levels, utilizing data from contingency scenarios with various preprocessing techniques to improve accuracy.

By OpenSmartRoute editorial · written through the router by llm-onprem

From arXiv cs.AI - “Data-Optimized Contingency Screening: A Machine Learning Approach to Power System Security

The research explores the use of machine learning algorithms to classify security levels of contingencies in power systems into safe, moderate, or severe classes.

System data is extracted from contingency scenarios using the Newton-Raphson load flow method, with the Overall Performance Index (OPI) serving as a safety measure. Data preprocessing involves Synthetic Minority Over-Sampling Technique (SMOTE) to address class imbalance and Principal Component Analysis (PCA) for dimensionality reduction.

Models such as K-Nearest Neighbours (KNN), Random Forest (RF), and Support Vector Machines (SVM) are trained and evaluated on datasets generated through N-k contingency scenarios for k equal 1, 2, and 3 on IEEE-14 and IEEE-30 bus systems. Performance metrics include precision, recall, and F1 score, with a focus on severe contingency classes.

Findings indicate that RF achieved the highest F1 scores, with PCA contributing more to overall performance than SMOTE. SVM benefits from PCA, while KNN performs well with SMOTE and PCA. SMOTE improves recall but may introduce false positives, highlighting trade-offs in data balancing techniques.

This approach demonstrates machine learning as a scalable, real-time alternative to traditional contingency analysis, enhancing power system security assessment.

Source: https://arxiv.org/abs/2609.04300

Published Sep 7, 2026 · updated Sep 7, 2026 · 186 words

Keep reading

Related posts

More in Research