کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

Cluster-Based Multi-Model Learning for Enhanced Accuracy of Optimal Power Flow Solution

نویسندگان :

Reza Mohammad Yousefi¹ Ahmad Salehi Dobakhshari²

1- دانشگاه گیلان 2- دانشگاه گیلان

کلمات کلیدی :

Clustering،Machine Learning،Multi-Model Approach،Optimal Power Flow،Power Systems Optimization

چکیده :

The Optimal Power Flow (OPF) problem is fundamental for optimizing the operation of modern power systems, which are becoming increasingly complex due to various dynamic factors. Traditional optimization techniques, while effective, are often slow and face challenges related to limited scalability in large and varying systems. Standard machine learning (ML) methods, while faster, often rely on a single, large model that struggles to generalize across diverse operational conditions. In this paper, we introduce a cluster-based multi-model learning framework, which divides the OPF solution space into distinct operational regimes through data-driven clustering methods. By training specialized machine learning models for each cluster, the framework adapts more effectively to varying system scenarios. This improves prediction accuracy and adaptability by tailoring models to specific operational regimes. Simulation results on standard IEEE test systems with more than 1000 buses demonstrate that the proposed method significantly enhances accuracy and provides a more scalable solution for real-time grid optimization compared to both conventional ML-based single-model techniques and traditional analytic optimization methods.