Two-layer multi-objective optimal sizing of electric-hydrogen energy storage with the integration of extreme scenario generation and preference-information decision-making

The increasing frequency of extreme events presents significant challenges to the security and reliability of electric-hydrogen energy systems, where traditional optimal sizing methods often fail to address supply–demand imbalance risks under typical conditions. To bridge this gap, in the present st...

Full description

Saved in:
Bibliographic Details
Main Authors: Zihan Sun, Jian Chen, Yang Chen, Wen Zhang, Tingting Zhang, Yicheng Zhang, Guangsheng Pan
Format: Article
Language:English
Published: Elsevier 2025-09-01
Series:International Journal of Electrical Power & Energy Systems
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S0142061525004399
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The increasing frequency of extreme events presents significant challenges to the security and reliability of electric-hydrogen energy systems, where traditional optimal sizing methods often fail to address supply–demand imbalance risks under typical conditions. To bridge this gap, in the present study, a comprehensive multi-objective optimization framework that integrates extreme scenario generation with preference-information decision-making is proposed. First, an integrated generative adversarial network-based method is developed to synthesize diverse scenarios while preserving long-duration temporal dependencies and accurately capturing wind-solar-load correlations. A two-layer multi-objective optimization model is subsequently established to coordinate short- and long-duration energy storage dispatch across both normal and extreme scenarios, thereby increasing the economic viability and reliability of the system. Furthermore, an angle preference-based multi-objective particle swarm optimization algorithm is introduced to incorporate the preferences of decision-makers and yield flexible and tailored optimal sizing solutions. The evaluation results demonstrate that the proposed integrated GAN scenario generation method reduced the training losses by 60.97% in comparison with other traditional methods. Additionally, the overall optimization costs for the extreme and normal scenarios were 1.08% and 12.5% lower, respectively, than those for the same scenarios without preference optimization. Finally, the effectiveness and scalability of the proposed methodology were validated within an extended electric-thermal-hydrogen system on historical extreme high-temperature scenario data.
ISSN:0142-0615