Scholarlink

<p>The optimization of sports training strategies requires balancing multiple conflicting objectives, including performance enhancement, injury prevention, and fatigue management. Traditional training planning methods rely heavily on coaches’ experience and fail to adapt dynamically to athletes’ changing physiological and psychological states. This paper introduces MOATS (Multi-Objective Adaptive Training Strategy), a novel framework that leverages multi-objective reinforcement learning (MORL) to generate personalized and adaptive training strategies. MOATS integrates a hierarchical state representation module that captures athletes’ multi-dimensional characteristics, a Pareto-based policy optimization mechanism that handles conflicting training objectives, and a temporal adaptation component that adjusts strategies based on real-time feedback. The framework employs a decomposition-based MORL algorithm combined with attention mechanisms to learn optimal trade-offs among competing objectives. Additionally, we introduce a physiological constraint module that ensures generated strategies comply with sports science principles. Experimental results on two real-world datasets demonstrate that MOATS achieves a 15.8% improvement in overall training effectiveness while reducing injury risk by 23.4% compared to existing methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The research on the generation method of sports training strategies based on multi-objective reinforcement learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月21日

查看全文
  • Min Zeng

摘要

The optimization of sports training strategies requires balancing multiple conflicting objectives, including performance enhancement, injury prevention, and fatigue management. Traditional training planning methods rely heavily on coaches’ experience and fail to adapt dynamically to athletes’ changing physiological and psychological states. This paper introduces MOATS (Multi-Objective Adaptive Training Strategy), a novel framework that leverages multi-objective reinforcement learning (MORL) to generate personalized and adaptive training strategies. MOATS integrates a hierarchical state representation module that captures athletes’ multi-dimensional characteristics, a Pareto-based policy optimization mechanism that handles conflicting training objectives, and a temporal adaptation component that adjusts strategies based on real-time feedback. The framework employs a decomposition-based MORL algorithm combined with attention mechanisms to learn optimal trade-offs among competing objectives. Additionally, we introduce a physiological constraint module that ensures generated strategies comply with sports science principles. Experimental results on two real-world datasets demonstrate that MOATS achieves a 15.8% improvement in overall training effectiveness while reducing injury risk by 23.4% compared to existing methods.