强化学习算法研究热点与发展趋势探析

An Analysis of Research Hotspots and Development Trends in Reinforcement Learning Algorithms

摘要: 为系统揭示2021年至2025年强化学习领域的研究态势与发展脉络，以中国知网收录的5099篇“强化学习”主题核心期刊论文为研究对象，采用文献计量法与可视化分析法，从文献时间分布、关键词共现与聚类、关键词突现等维度展开分析。研究发现：该领域文献呈快速扩张和稳定提质两阶段增长特征，前期聚焦基础算法改进与单一场景验证，后期转向细分场景深化与多技术融合；研究热点覆盖算法创新、多领域落地、技术协同与安全保障；前沿演进遵循技术融合到场景深化再到知识驱动的路径。研究可为科研人员把握领域热点、布局未来研究方向提供参考，为强化学习技术的产业化应用提供理论支撑。

Abstract: To systematically reveal the research status and development context of the reinforcement learning field from 2021 to 2025, this study takes 5099 core journal papers themed “reinforcement learning” included in China National Knowledge Infrastructure (CNKI) as the research objects, and conducts analysis from the dimensions of literature time distribution, keyword co-occurrence and clustering, and keyword burst detection by adopting the bibliometric method and visual analysis method. The results show that the literature in this field has two-stage growth characteristics of rapid expansion and steady quality improvement: the early stage focuses on basic algorithm improvement and single-scenario verification, while the later stage shifts to the deepening of subdivided scenarios and multi-technology integration. The research hotspots cover algorithm innovation, multi-field application, technical collaboration and security assurance. The frontier evolution follows the path from technology integration to scenario deepening and then to knowledge-driven development. This study can provide references for researchers to grasp the hotspots of the field and layout future research directions, as well as offer theoretical support for the industrial application of reinforcement learning technology.