Topology Optimization of Large-scale Offshore Wind Farm Collector Systems Based on Large Language Models

The optimization of the collector system topology is a core issue in the planning and construction of large-scale offshore wind farms (LSOWF), and it is inherently a complex mixed-integer optimization problem involving multiple constraints and objectives. To address this problem, this paper proposes a method for optimizing the collector system topology of LSOWF assisted by large language model (LLM). Firstly, groups of wind turbines (WT) with the assistance of LLM are clustered and the method of chain prompting is applied to help the LLM understand the optimization objectives. Then, the LLM is employed to divide LSOWFs into smaller areas. This reduces the dimensions of the optimization problem and enhances both solving speed and quality. Next, the optimization model for the collection system topology is constructed and a mixed-integer linear programming solver is utilized to achieve the optimal design for the LSOWF’s collector system topology. Finally, the method’s performance is validated using a LSOWF with 75 WTs. The simulation results show that, compared to traditional optimization techniques, the proposed method can be adopted to achieve more balanced clustering of WTs and generate a topology design that is economically optimal while considering power loss. The LLM demonstrates high effectiveness in assisting with the collector system topology optimization, providing a new approach for LSOWF collector system design optimization. © 2024 Science Press. All rights reserved.