基于检索增强的故障诊断知识问答模型

Retrieval Augmented Fault Diagnosis Knowledge Question Answering Model

大语言模型在日常对话、代码编写和情感识别等多个领域已表现出卓越的性能。然而,大语言模型在进行故障诊断领域的问答任务时会出现幻觉问题。针对此问题,提出了一种基于检索增强的故障诊断知识问答模型。该模型结合了自建的故障诊断领域知识库,显著提升了大语言模型在故障诊断领域的知识问答能力,缓解了大语言模型在进行故障诊断领域的知识问答任务时出现的幻觉问题。在处理故障诊断知识问答和故障类型判断任务时,该模型的表现远超常规的大语言模型。

Large Language Models(LLMs)have demonstrated exceptional performance in various domains such as everyday conversations,code writing,and emotion recognition.However,LLMs exhibit hallucination problems when performing question-answering tasks in the field of fault diagnosis.To address this issue,we propose a retrieval-enhanced fault diagnosis knowledge question-answering model.This model incorporates a self-constructed knowledge base specific to the fault diagnosis domain,significantly improving the LLMs question-answering capabilities in this field and alleviating the hallucination problems associated with fault diagnosis tasks.In handling fault diagnosis knowledge question-answering and fault type determination tasks,this model greatly outperforms conventional LLMs.