کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

A Hybrid Forecast-Evolutionary Framework with Random Forest for Optimal PV and Capacitor Bank Placement in Microgrids

نویسندگان :

Sepehr Serajian¹ Mohammad Hassan Nazari² Behrooz Vahidi³ Seyed Hossein Hosseinian⁴

1- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران) 2- Niroo Research Institute (NRI) 3- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران) 4- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران)

کلمات کلیدی :

Voltage Stability،Microgrids،photovoltaic،Machine Learning،Capacitor banks

چکیده :

The increasing penetration of photovoltaic (PV) systems in distribution networks has amplified challenges such as higher losses and voltage instability. While many optimization studies assume perfect foresight of load and generation, this is impractical in real-world operation. This paper addresses this gap by combining short-term forecasts from Random Forest (RF), Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN1D) models with the Evolutionary Metaheuristic Based on Torrent (EM-BT) algorithm. Load and irradiance were forecasted at horizons of 1, 6, and 12 hours and used to optimize PV siting and capacitor bank allocation on the IEEE-33 bus feeder. Results show that forecast-informed EM-BT reduced feeder losses by about 21% and raised minimum voltages from 0.90 pu to 0.96 pu, slightly outperforming oracle optimization. The framework demonstrates practical value for both planning and operations, enabling smarter and more resilient distribution networks.