安全、健康和环境2026，Vol.26Issue(6)：15-20,6.DOI:10.3969/j.issn.1672-7932.2026.06.002
基于RAG与DeepSeek的HAZOP智能辅助系统优化
Optimizing HAZOP Analysis via an Intelligent Auxiliary System Based on RAG and DeepSeek
刘伟1
作者信息
- 1. 中国石油化工集团有限公司健康安全环保管理部,北京 100020
- 折叠
摘要
Abstract
To address the issues of low efficiency,poor consistency,and heavy reliance on expert experi-ence in traditional Hazard and Operability(HAZOP)analysis in chemical enterprises,an intelligent auxiliary optimization method based on Retrieval-Augmented Generation(RAG)technology and the DeepSeek large lan-guage model was proposed.Specifically,this method constructed a structured vector database integrating multi-source heterogeneous safety knowledge and utilized the domain-knowledge-enhanced model to achieve the auto-mated review and in-depth supplementation of HAZOP analysis results.A case study demonstrated that the sys-tem effectively identified the omissions of typical hazards and insufficient cause identification within the traditional analysis,subsequently generating targeted optimization recommendations.These findings offer an innovative and intelligent tool for enhancing the completeness and reliability of HAZOP analysis in the chemical industry.关键词
检索增强生成/化工安全/HAZOP分析/安全知识库/DeepSeek大模型Key words
Retrieval-Augmented Generation(RAG)/chemical safety/HAZOP analysis/safety knowl-edge database/DeepSeek model分类
资源环境引用本文复制引用
刘伟..基于RAG与DeepSeek的HAZOP智能辅助系统优化[J].安全、健康和环境,2026,26(6):15-20,6.
