کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

Data-Driven Intelligent Islanding Detection in Inverter Based Distributed Generation Systems Using Load Characteristics

نویسندگان :

Masoumeh Seyedi¹ Fereshteh Poorahangaryan²

1- دانشگاه غیردولتی آیندگان 2- دانشگاه غیردولتی آیندگان

کلمات کلیدی :

Islanding Detection،Inverter-Based Distributed Generation (DG)،Machine-Learning،Gaussian Mixture Model (GMM)

چکیده :

This study presents an advanced islanding detection strategy for inverter interfaced distributed generation, based on a learning driven characterization of load behavior. Unlike conventional frequency-based relays, the method introduces a controlled reactive power imbalance to induce measurable frequency deviations, adaptively tuned according to load properties such as resonant frequency and quality factor. Load signatures are analyzed via a hybrid machine learning framework, where Gaussian Mixture Models (GMM) cluster operating conditions and a regression estimator infers load coefficients, enabling an optimal d-q axis current modulation to enhance detection under islanded conditions. The approach is validated across diverse load scenarios, including IEEE 1547-2018 and IEEE 929-2000 compliant cases. Simulation results show consistent islanding detection within 30 ms, reduced non-detection zones, and mitigated transient voltage and frequency disturbances. By integrating data-driven load assessment with adaptive control, the proposed framework improves detection reliability and responsiveness without additional hardware, offering a robust solution for integrating inverter-based renewable energy resources into distribution networks.