关键词: 故障诊断, 提示学习, 信息抽取, 知识图谱

Abstract: With the continuous advancement of artificial intelligence and information technology,fault diagnosis for manufacturing equipment has increasingly adopted new diagnostic methods such as machine learning and knowledge graphs. In particular,the knowledge graph for manufacturing equipment fault diagnosis integrates vast amounts of fault-related knowledge,enabling the inference of fault causes and solutions based on specific fault issues. This provides a novel approach for rapid fault diagnosis of manufacturing equipment. However,traditional methods for constructing knowledge graphs require substantial manual effort and are both complex and cumbersome. To address this issue,it proposes a knowledge graph construction framework based on prompt learning. By leveraging large language models GPT,the schema layer of the manufacturing equipment fault knowledge graph is defined. Furthermore,an information extraction model,PNUIE,based on prompt learning is introduced to identify and extract triplet information from textual data,thereby constructing a fault diagnosis knowledge graph for manufacturing equipment. Experimental results demonstrate that the proposed method effectively achieves efficient extraction of fault knowledge,outperforming existing methods in terms of performance.

Key words: fault diagnosis, prompt learning, information extraction, knowledge graph