铁路通信信号工程技术2026，Vol.23Issue(8)：22-27,6.DOI:10.3969/j.issn.1673-4440.2026.08.003
基于大语言模型的ZPW-2000A轨道电路智能诊断系统研究
Research on Intelligent Diagnosis System for ZPW-2000A Track Circuit Based on Large Language Model
摘要
Abstract
This study proposes an intelligent diagnosis system for ZPW-2000A track circuit based on Large Language Model(LLM),to address the problem of low diagnosis accuracy for complex faults in ZPW-2000A track circuit caused by the reliance on manual experience and static rule libraries for on-site fault diagnosis.By integrating multi-modal knowledge base construction methods with retrieval augmented generation techniques,this system achieves deep semantic correlation and dynamic inference of track circuit fault features.Based on the idea of multi-level hierarchical and multi-step chain of thought reasoning,a mechanism of"fault category recognition-fault area focusing-fault equipment diagnosis"is constructed to achieve accurate identification of track circuit faults,provide professional analysis and troubleshooting suggestions,and enable dynamic generation of fault handling reports,thereby providing technical support for real-time decision-making of operation and maintenance personnel.关键词
ZPW-2000A/大语言模型(LLM)/故障诊断Key words
ZPW-2000A/Large Language Model(LLM)/fault diagnosis分类
信息技术与安全科学引用本文复制引用
李茜钰..基于大语言模型的ZPW-2000A轨道电路智能诊断系统研究[J].铁路通信信号工程技术,2026,23(8):22-27,6.基金项目
中国铁路通信信号股份有限公司项目课题(2025K07) （2025K07）
