首页|期刊导航|铁路通信信号工程技术|基于大语言模型的ZPW-2000A轨道电路智能诊断系统研究

基于大语言模型的ZPW-2000A轨道电路智能诊断系统研究OA

Research on Intelligent Diagnosis System for ZPW-2000A Track Circuit Based on Large Language Model

中文摘要英文摘要

针对ZPW-2000A型轨道电路现场故障诊断依赖人工经验及静态规则库,导致的复杂故障诊断率低的问题,提出一种基于大语言模型(Large Language Model,LLM)的ZPW-2000A轨道电路智能诊断系统.系统通过融合多模态知识库构建方法与检索增强生成技术,实现轨道电路故障特征的深度语义关联与动态推理.采用多级分层、多步链式推理的思想构建"故障类别识别-故障区域聚焦-故障设备诊断"机制,实现轨道电路故障的精准识别,给予专业分析和故障排查建议,动态生成故障处理报告,为运维人员的实时决策提供技术支持.

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.

李茜钰

北京全路通信信号研究设计院集团有限公司,北京 100070||列车自主运行智能控制铁路行业工程研究中心,北京 100070

信息技术与安全科学

ZPW-2000A大语言模型(LLM)故障诊断

ZPW-2000ALarge Language Model(LLM)fault diagnosis

《铁路通信信号工程技术》 2026 (8)

22-27,6

中国铁路通信信号股份有限公司项目课题(2025K07)

10.3969/j.issn.1673-4440.2026.08.003

评论