基于大模型多智能体协同的油气销售企业智能系统运维研究  

Research on AIOps of oil and gas sales enterprises based on LLMs and MAS collaboration

作　　者：曹路 孙世亮 CAO Lu;SUN Shiliang(PetroChina Marketing Company,Beijing 100007,China;PetroChina Heilongjiang Marketing Company,Harbin 150010,China)

机构地区：[1]中国石油天然气股份有限公司销售分公司,北京100007 [2]中国石油天然气股份有限公司黑龙江销售分公司,哈尔滨150010

出　　处：《车用能源储运销技术》2024年第6期45-53,共9页

摘　　要：随着油气销售行业数字化转型进程的不断深入,数据规模的快速增长与业务复杂性的不断提高给企业系统运维工作带来了巨大挑战。大模型与多智能体系统以其强大的数据分析和协同能力,为系统运维提供了全新的技术手段与应用思路。本文从油气销售企业系统运维现状分析出发,提出基于大模型和多智能体协同的智能系统运维方案,详细阐述其系统架构、智能体组成及协同机制,并结合实际场景探讨方案的典型应用与预期效果。研究结果表明,所述方案能够显著提升运维效率与可靠性,降低运维成本,为企业数字化转型和系统稳定运行提供重要支撑。With the deepening of digital transformation in the oil and gas sales industry,the exponential growth in data volume and increasing business complexity have posed significant challenges to enterprise system operation and maintenance(O&M).Large language models(LLMs)and multi-agent systems(MAS),with their robust data analysis and collaboration capabilities,offer novel technical solutions and application paradigms for AIOps.This paper analyzes the current state of system operation and maintenance in oil and gas sales enterprises and proposes an intelligent system operation and maintenance framework based on LLMs and MAS cooperation.It details the system architecture,agent composition,and cooperation mechanisms,while also discussing typical applications and anticipated outcomes through real-world scenarios.Research findings indicate that the proposed framework can substantially enhance O&M efficiency and reliability,reduce costs,and provide critical support for enterprise digital transformation and system stability.

关 键 词：大模型 智能体协同 智能系统运维 油气销售 人工智能 

分 类 号：F426[经济管理—产业经济]