基于MLP和LSTM多因素网球比赛中动量波动胜负趋势模型研究  

MLP and LSTM Based Modeling of Momentum-fluctuating Win/Loss Trend in Multi-factor Tennis Matches

作　　者：贠欣欣 冯孝周[1] 邢润强 任笑笑 时华[1] 胡凯 YUN Xinxin;FENG Xiaozhou;XING Runqiang;REN Xiaoxiao;SHI Hua;HU Kai(School of Science,Xi'an Technological University,Xi'an 710021,China;University of Chinese Academy of Sciences,Beijing 101400,China;Xi'an Institute of Optics and Precision Mechanics,Chinese Academy of Sciences,Xi'an 710119,China)

机构地区：[1]西安工业大学基础学院,陕西西安710021 [2]中国科学院大学,北京101400 [3]中国科学院西安光学精密机械研究所,陕西西安710119

出　　处：《内蒙古师范大学学报(自然科学版)》2025年第2期188-197,共10页Journal of Inner Mongolia Normal University(Natural Science Edition)

摘　　要：通过深入分析部分温布尔登网球公开赛男子组决赛的比赛数据,探讨网球比赛中势头变化的现象,并创建了一套新的综合胜率指标和发球方优势指标,利用这些指标构建势头模型,预测比赛的胜负并制定更科学合理的战术和训练策略。首先,对数据进行处理并对特征进行筛选和创建,构建Momentum模型评价指标体系;然后,通过分析比赛中的波动与连续成功之间的关系,发现相关性显著;最后,综合考虑多方面因素,提出一种基于多层感知机模型与长短期记忆网络的组合预测模型,结合多层感知机(multilayer perceptron,MLP)的特征学习能力强与长短期记忆网络(long short-term memory,LSTM)预测精度高的优点,对比赛波动趋势进行高效精准预测,构建M‐LSTM模型预测比赛动态变化和胜负趋势,并使用Shap值方法可视化特征重要性,结果显示模型可以很好捕捉比赛的大体趋势,预测最终结果,对相关科学研究和网球运动的发展具有重要的参考意义。This study delved into the phenomenon of momentum change in tennis matches by analyzing in depth the match data of some Wimbledon men's finals.Further,it created a new set of comprehensive win rate indicators and serve-side dominance indicators and used these indicators to build a momentum model,so as to predict the winners of the matches and formulate more scientific and reasonable tactics and training strategies.First,the momentum model evaluation index system was constructed after data processing and screening and creation of features.Then,analysis was made on the relationship between fluctuation and consecutive success in the matches,revealing a significant correlation.Finally,considering multiple factors,the study proposed a combined prediction model based on a multilayer perceptron(MLP)model and a long short-term memory(LSTM)network.Combining the strong feature learning ability of MLP and high prediction accuracy of the LSTM network,the proposed model realized efficient and accurate prediction of the match fluctuation trend.Besides building the M-LSTM model to predict the dynamic changes of the matches and the trend of winning and losing,this study also used the Shap value method to visualize feature importance.The results showed that the model can capture the general trend of the matches and predict the final results well,which provided significant reference for related research and tennis development.

关 键 词：网球运动 深度学习 综合胜率指标 M‐LSTM模型 Shap值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]