【摘要】 RoboCup 2D是一项基于虚拟仿真环境的国际机器人足球竞赛，射门优化是提升智能体对抗胜率的核心手段，在智能决策与战术协同等领域具有重要研究价值。针对传统射门优化方法将射门视为单一动作、忽视不同射门形式差异的问题，本文基于现代足球理论，提出一种以关键传球射门为核心的优化分析方法。首先明确定义了仿真环境下的关键传球射门，并通过实际比赛数据统计分析证明，该方式的成功率较非关键传球射门提升10.9个百分点，验证了其战略价值。为进一步探究其影响因素，构建了基于XGBoost的分类模型，并结合SHAP进行可解释分析。实验结果显示，XGBoost模型在测试集上的分类准确率达92.1%，优于其他对比模型；SHAP可解释性分析结果表明，射门区传球控球率与传球次数是影响关键传球射门形成的主要特征。更多还原

【Abstract】 RoboCup 2D is an international virtual simulated robot soccer competition where shot optimization is crucial for improving agent win rates in competitive scenarios and has significant research value in intelligent decision-making and tactical coordination. Addressing the limitation of traditional methods that treat shooting as a uniform action and overlook differences in shooting styles, this paper proposes an optimized analytical approach focusing on key-pass shots based on modern football theory.Firstly, key-pass shots are explicitly defined in the simulation environment, and statistical analysis of real-game data shows they increase the success rate of shots by 10.9% compared to non-key-pass shots, confirming their strategic importance. To further explore the influencing factors, an XGBoost-based classification model combined with SHAP for interpretability analysis reveals that passing possession and pass frequency in the shooting area are the main factors influencing key-pass shot formation. Experimental results show that the model achieves 92.1% classification accuracy on the test set, outperforming other benchmark models.更多还原