金融与风险管理
Finance and Risk Management
- 主办单位:東方陽光出版社有限公司
- ISSN:3080-9320(P)
- ISSN:3080-9339(O)
- 期刊分类:经济管理
- 出版周期:月刊
- 投稿量:0
- 浏览量:122
相关文章
暂无数据
烟草商业消防防控体系转型研究——基于大语言模型的低成本可验证文本治理路径
Research on the Transformation of Fire Prevention and Control System in Tobacco Commercial Enterprises —A Low-Cost and Verifiable Text Governance Path Based on Large Language Models
发布时间：2026-07-21
作者: 王玮 :贵州省烟草公司贵阳市公司 贵州贵阳; WANG Wei :Guiyang Municipal Company of Guizhou Tobacco Company, Guiyang;
摘要: 烟草商业企业消防防控长期面临知识管理碎片化、隐患排查标准化程度低、数字化建设成本高等结构性矛盾。与传统智慧消防依赖物联网硬件的高投入路径不同，本文提出基于大语言模型（LLM）的文本智能化转型方案，通过构建领域知识增强的检索生成系统，实现消防法规的语义化检索、隐患描述的辅助标准化以及培训内容的个性化生成。研究结合智慧消防、烟草仓储火灾研究及大语言模型可靠性相关文献，对开源大模型在烟草商业场景下的技术适配性进行分析，设计了"本地轻量化部署+提示工程优化"的低成本实施框架，并通过对典型地市级公司消防管理流程的解构，讨论了该方案在现有IT基础设施下的可行性。研究认为，基于文本数据治理的智能化路径有助于缓解基层单位数字化转型中的成本与能力约束，为传统行业安全管理提供了一种成本相对可控、可分阶段推进的转型思路。
Abstract: Tobacco commercial enterprises have long faced structural contradictions in fire prevention and control, including fragmented knowledge management, low standardization of hazard identification, and high digitalization costs. Unlike traditional smart fire protection approaches that rely heavily on IoT hardware investment, this paper proposes a text-based intelligent transformation scheme leveraging large language models (LLMs). By constructing a domain knowledge-enhanced retrieval-augmented generation (RAG) system, the scheme enables semantic retrieval of fire safety regulations, assisted standardization of hazard descriptions, and personalized generation of training content. Drawing on literature related to smart fire protection, tobacco warehouse fire research, and LLM reliability, the study analyzes the technical adaptability of open-source LLMs in tobacco commercial scenarios and designs a low-cost implementation framework featuring "local lightweight deployment + prompt engineering optimization." Through deconstructing the fire management workflow of a typical prefecture-level tobacco company, the feasibility of this solution under existing IT infrastructure is discussed. The study suggests that the text-based intelligent governance pathway helps alleviate the cost and capacity constraints faced by grassroots units in digital transformation, offering a relatively controllable-cost and phased-advancement approach for safety management in traditional industries.
关键词: 大语言模型；烟草商业；消防防控；知识管理；低成本数字化
Keywords: large language model; tobacco commerce; fire prevention and control; knowledge management; low-cost digitalization