大模型思维链与指挥决策链融合构建技术研究
CSTR:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

TP18;TP391;E0-059

基金项目:


Research on technology of integrating chain of thought in LLMs with chain of command and decision
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对大模型(LLM)在指挥决策中存在的推理能力不足和“幻觉”问题, 通过系统梳理大模型思维链(CoT)和智能体指挥决策链的技术范式, 提出一种架构与思维双重维度的融合方法, 并借助工程实践验证其有效性. 研究表明, 思维链与指挥决策链的融合可增强大模型在指挥决策中的认知推理能力, 推动该类任务由经验依赖向认知驱动转型. 最后, 讨论指挥决策大模型和指挥决策智能体的未来发展趋势, 并提出可能的解决思路和研究方向.

    Abstract:

    Catering to addressing the issues of insufficient reasoning capabilities of LLMs (large language models) and “hallucination” phenomenon in LLMs in the domain of the command and control (C2), this paper systematically reviews the technical paradigms of the chain of thought (CoT) in LLMs and chain of command and decision (CoCD) in AI agents. A method of integrating CoT in LLMs and CoCD from architectural and reasoning perspectives is proposed and validated through engineering practices. The study demonstrates that integrating CoT in LLMs with CoCD can enhance the cognitive reasoning capabilities of LLMs in command and decision, promoting a shift from experience-dependent to cognitively-driven approaches for such tasks. Finally, the paper discusses future trends in LLMs and AI agents in the domain of the command and control, suggesting potential solutions and research directions.

    参考文献
    相似文献
    引证文献
引用本文

郑雅芝,崔翛龙.大模型思维链与指挥决策链融合构建技术研究[J].控制与决策,2026,41(2):432-444

复制
相关视频

分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2025-05-23
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2026-01-17
  • 出版日期:
文章二维码