Robust Optimal Operation of Smart Microgrid Considering Source–Load Uncertainty

Uloženo v:

Podrobná bibliografie
Vydáno v:	Processes vol. 13, no. 8 (2025), p. 2458-2484
Hlavní autor:	Qiu Zejian
Další autoři:	Zhu Zhuowen, Yu, Lili, Han Zhanyuan, Shao Weitao, Zhang, Kuan, Ma Yinfeng
Vydáno:	MDPI AG
Témata:	Load Accuracy Distributed generation Voltage Optimization Signal processing Power flow Closed loops Renewable energy Long short-term memory Statistical analysis Uncertainty Probability distribution Climate change Prediction models Robustness Wind power Electric potential Confidence intervals Neural networks Relaxation method (mathematics) Renewable resources Error reduction Mixed integer Alternating current Alternative energy sources Decision making Energy distribution
On-line přístup:	Citation/Abstract Full Text + Graphics Full Text - PDF
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!

Popis
Abstrakt:	The uncertainties arising from high renewable energy penetration on both the generation and demand sides pose significant challenges to distribution network security. Smart microgrids are considered an effective way to solve this problem. Existing studies exhibit limitations in prediction accuracy, Alternating Current (AC) power flow modeling, and integration with optimization frameworks. This paper proposes a closed-loop technical framework combining high-confidence interval prediction, second-order cone convex relaxation, and robust optimization to facilitate renewable energy integration in distribution networks via smart microgrid technology. First, a hybrid prediction model integrating Variational Mode Decomposition (VMD), Long Short-Term Memory (LSTM), and Quantile Regression (QR) is designed to extract multi-frequency characteristics of time-series data, generating adaptive prediction intervals that accommodate individualized decision-making preferences. Second, a second-order cone relaxation method transforms the AC power flow optimization problem into a mixed-integer second-order cone programming (MISOCP) model. Finally, a robust optimization method considering source–load uncertainties is developed. Case studies demonstrate that the proposed approach reduces prediction errors by 21.15%, decreases node voltage fluctuations by 16.71%, and reduces voltage deviation at maximum offset nodes by 17.36%. This framework significantly mitigates voltage violation risks in distribution networks with large-scale grid-connected photovoltaic systems.
ISSN:	2227-9717
DOI:	10.3390/pr13082458
Zdroj:	Materials Science Database