.png){kind=icon}
.png){kind=icon}
Return
Research on wind turbine fault root cause localization method integrating large language models and causal hypergraphs
.png){kind=icon}
PRE
AI
Y
DOI：10.1016/j.compeleceng.2026.111335.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Wind turbine fault root cause localization is essential for ensuring safe operation of wind farms. However, existing causal hypergraph methods struggle to balance interpretability of reasoning processes with effective utilization of domain knowledge when modeling high-order causal interactions among multiple components. This paper proposes a large language model (LLM) enhanced causal hypergraph reasoning framework for wind turbine fault root cause localization. The framework first employs LLMs to extract causal relationship knowledge from technical documents and fault cases of wind turbines, then formalizes this knowledge as constraint conditions for structure learning, forming a knowledge-guided causal hypergraph structure learning method. Subsequently, the framework designs a symbolic representation format for causal propagation paths and transforms the reasoning process into natural language diagnostic reports through LLMs, forming a semantic-enhanced causal reasoning chain generation method. Experimental results on the Case Western Reserve University (CWRU) bearing dataset, the Paderborn University bearing dataset, and a wind turbine supervisory control and data acquisition (SCADA) dataset demonstrate that the proposed method achieves 92.31%, 89.23%, and 89.78% root cause localization accuracy respectively, with reasoning chain interpretability scores of 4.20 out of 5.
Journal
C
IF:
4.9
Papers:
6.7K
Citations:
1.3W
Organization
No organization information available
Cited Papers