.png){kind=icon}
.png){kind=icon}
Return
Integrating domain knowledge and large language models for automatic generation of function block-based PLC logic in maritime systems
.png){kind=icon}
PRE
AI
H
J
I
DOI：10.1016/j.jss.2026.112991.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A domain-aware LLM–RAG framework is proposed to automatically generate Function Block–based PLC logic for maritime control systems without model fine-tuning. • Graph-based relational knowledge—FB selection/cooperation rules and I/O group/symmetry structures—is extracted and injected to enhance FB planning and instance estimation. • Natural-language transformation of I/O descriptions and redundancy-filtered FB embeddings significantly improve RAG retrieval performance in real ship datasets. • Structured planning combined with relationship-aware reasoning improves FB-type F1 scores and raises instance estimation accuracy to over 88% in large LLMs. • A rule-based IEC 61131-3 code generator ensures syntactic and structural validity, enabling deployable FB-based logic generation for industrial automation workflows.
Journal
IF:
4.1
Papers:
5.4K
Citations:
8.4K
Organization
Cited Papers