摘要
传球动作是RoboCup仿真足球比赛得以进行的纽带,进球并赢得比赛是球队的最终目标,分析传球与比赛胜负的关系,采用数据挖掘的思想,用C语言程序解析仿真比赛日志文件的方法来收集所需的传球数据,基于距离将传球分为5种类型,以5种类型的传球作为解释变量,以比分作为因变量,采用偏最小二乘法搭建数学模型,用SIMCA-P软件进行仿真实验,并采用相关图形进行分析与说明。在包含解释变量72.8%、因变量74.4%信息量的情况下,5个解释变量对因变量的投影重要性指标值分别为0.081 14,0.996 66,1.028 9,1.088 06,1.325 73。实验结果表明,对传球来说,长传球对比赛胜负的影响最大。
Passing ball action runs through the RoboCup simulation games,both teams spare no effort to win the game.The paper studies the connection between passing ball action and the game deeply. It puts forward to adopt the idea of data mining,analyzes games＇ log files by C language program in order to collect the required passing ball data,divides passing ball into5 types which are seen as independent variables and see score as dependent variable,then establishes mathematical model combining with Partial Least Square（PLS）. A few relevant figures are used to analyze and verify the experimental result which comes from SIMCA-P. The result shows that with72.8% of independent variables information and74.4% of dependent variable information,the VIP values of5 independent variables to dependent variable are as follows：0. 081 14,0. 996 66,1. 028 9,1.088 06,1.325 73. After linking theoretical result with practical scene,it is concluded that long pass plays a major role in a game for passing ball.
出处
《计算机工程》
CAS
CSCD
2014年第9期275-279,共5页
Computer Engineering
基金
安徽高校省级自然科学研究基金资助项目(KJ2011A039)
关键词
多智能体系统
日志文件
数学建模
偏最小二乘法
回归分析
多重共线性
Multi-Agent System（MAS）
log file
mathematical modeling
Partial Least Square（PLS）
regression analysis
multi-collinearity
