基于人工神经网络的蹦床运动员竞技能力结构评价模型的构建与应用
A Research on Competitive Ability Structure Model of Trampoline Athletes Based on Artificial Neural Network
  
DOI:10.12064/ssr.2021090701
中文关键词:蹦床  竞技能力结构  人工神经网络
英文关键词:trampoline  structural model of athletic ability  artificial neural network
基金项目:上海市体育局科技综合计划项目(16Z013)
作者单位
王乐军 同济大学 体育教学部,上海 200092 
王钰婷 上海体育科学研究所(上海市反兴奋剂中心),上海 200030 
龚铭新 同济大学 体育教学部,上海 200092 
邹凝祥 同济大学 体育教学部,上海 200092 
俞华 上海市竞技体育训练管理中心 体操中心,上海 202162 
章晓菁 上海市竞技体育训练管理中心 体操中心,上海 202162 
叶晓东 上海市竞技体育训练管理中心 体操中心,上海 202162 
摘要点击次数: 244
全文下载次数: 193
中文摘要:
      目的:建立蹦床运动员竞技能力结构评价指标体系,在此基础上构建基于人工神经网络的蹦床运动员竞技能力结构评价模型,为蹦床运动员竞技能力结构的个性化诊断及针对性训练提供参考。方法:以上海市体操运动中心16名蹦床运动员为研究对象,对受试者进行3次跨度6个月以上的初选指标测试。基于因子分析建立蹦床运动员竞技能力结构评价指标体系。在此基础上以竞技能力结构评价指标为自变量,运动员成绩为因变量,构建运动员竞技能力结构的人工神经网络评价模型,并开发运动员竞技能力结构评价系统。结果:蹦床运动员竞技能力结构指标体系由身体形态、身体素质、专项技术和心理素质4个维度构成,包括腿长、腿长/身高比、纵跳高度、原地立臂角度、60 s悬垂举腿、立卧撑、网上腾空高度、空跳高度/原地纵跳高度比、着网瞬间立臂角度、30次空跳高度下降率、状态焦虑水平和特质焦虑水平共12个指标。所构建的Elman人工神经网络模型由12个输入节点、9个隐含层节点和1个输出层节点组成,模型预测精度在95.87%~99.37%,平均预测精度高达97.66%。结论:构建了基于人工神经网络的蹦床运动员竞技能力结构评价模型,模型具有较好的预测精度。在训练中,可应用人工神经网络对竞技能力结构进行评价,动态获知竞技能力结构改变对总体运动成绩的影响作用。该研究对于蹦床运动员竞技能力结构的综合评价和针对性训练可提供科学性指导意见。
英文摘要:
      This study aims to establish the evaluation indices system and competitive ability structure model of trampoline athletes and thus to provide reference for personalized diagnosis and sport training of trampoline athletes. Sixteen trampoline athletes participated in the test of preliminary evaluation index for three times with six-month interval for two consecutive tests. The final evaluation indices were determined by the factor analysis of test data and the competitive ability structure model of trampoline athletes based on artificial neural network was established, in which the evaluation indices were taken as independent variables while the sport performance was taken as dependent variable. It finds that the competitive ability evaluation indices of trampoline athletes can be divided into four parts of physical appearance, physical ability, special ability, psychology quality and consisted of leg length, ratio of leg length to height, vertical jump height, shoulder flexibility angle, tuck hang times in 60-second, standing-to-push-ups times in 30-second, height of arch on trampoline, ratio of arch height on trampoline to vertical jump height, shoulder flexibility angle in the landing moment, decrease ratio of height of arch during 30-times vertical jump on trampoline, state anxiety level and trait anxiety level. The Elman artificial neural network was established with twelve input nodes, nine hidden layer nodes and one output node. The prediction accuracy of the model was between 95.87% and 99.37%, and the average prediction accuracy is as high as 97.66%. In conclusion, a competitive ability structure model of trampoline athletes based on artificial neural network was constructed. The model has high prediction accuracy and can be used to the evaluation and training improvement program of the competitive ability structure of trampoline athletes.
查看全文  查看/发表评论  下载PDF阅读器
关闭