知识增强大模型在信息系统故障分析中的应用研究

Research on the application of knowledge enhanced LLM in fault analysis of information system

典型的信息系统故障分析与处理任务高度依赖经验数据。将大语言模型应用于这一专业任务可以显著提高故障分析处理的效率与准确性。但通用大模型在专业知识问答任务中存在准确性低、知识不足且陈旧的问题。为此,设计了基于微调大模型与知识增强框架的故障分析系统。系统通过LoRA方法基于故障分析知识数据微调ChatGLM2-6B模型,通过LangChain框架融合微调大模型与故障分析知识库。对比实验表明,该系统在故障分析主观问答任务中具有更好的性能表现,提高了信息系统故障分析的准确性与自动化水平。

The typical fault analysis and processing tasks of information system are highly dependent on empirical data.Apply-ing large language model(LLM)to this professional task can significantly improve the efficiency and accuracy of fault analysis and processing.However,there are issues such as low accuracy,insufficient and outdated knowledge when using LLM in the profes-sional knowledge answering tasks.Therefore,a fault analysis system based on fine-tuning LLM and knowledge enhanced framework is designed.The system uses LoRA to fine-tune ChatGLM2-6B on the knowledge data of fault analysis,and integrates LLM with the knowledge base of fault analysis through LangChain.Comparative experiments show that the system has better performance in the subjective question-and-answer task of fault analysis,and improves the accuracy and automation level of fault analysis in informa-tion system.