基于RBF网络的足球点球轨迹预测
    点此下载全文
引用本文:周帅,张云飞,郑永权,许大炜.基于RBF网络的足球点球轨迹预测[J].计算技术与自动化,2024,(1):25-31
摘要点击次数: 58
全文下载次数: 0
作者单位
周帅,张云飞,郑永权,许大炜 (西安交通大学 城市学院陕西 西安 710018) 
中文摘要:基于计算机视觉线性化轨迹预测模型在预测足球轨迹时,只能保证局部稳定性,存在轨迹跟踪局部稳定性问题和parking问题,提出了基于RBF网络的足球点球轨迹预测方法。建立足球运动状态传感信号解析模型,计算足球飞行地心重力、空气阻力、空气浮力、自身旋转时产生的马格努斯力的参数。建立足球飞行过程中的飞行受力解析模型,鉴于多参数模型复杂度过高,产生parking问题。利用RBF网络模型简化能力,建立飞行轨迹预测模型,结合并行滤波控制器,融合以上所有信息,在已知视觉概率计算的基础上,完成足球飞行轨迹的状态估计,并将其误差的协方差计算作为滤波控制的输入值,从而得到所有方差、均值的数据。最后获得足球在任意时刻的运动状态函数,完成预测。实验结果显示,该方法的足球运行轨迹吻合度为12 mm,且落点距离标准差最大仅为0.0412 m,因此,该预测方法能够得到精度更高的预测数据。
中文关键词:RBF网络  足球点球  并行滤波  控制器  运行轨迹预测  传感参数
 
Prediction of Football Penalty Kick Trajectory Based on RBF Network
Abstract:Based on the linearized trajectory prediction model, when predicting the football trajectory, it can only guarantee the local stability, and there are local stability problems and parking problems in trajectory tracking. This paper proposes a method for predicting the trajectory of football penalty kick based on RBF network. The analytical model of football motion state sensing signal is established, and the parameters of gravity, air resistance, air buoyancy and Magnus force generated when the football is flying are collected. An analytical model of flight force in football flight is established. in view of the high complexity of the multi parameter model, the parking problem arises. Using the RBF network model simplification ability, a flight trajectory prediction model is established. Combining with the parallel filter controller, all the above information is fused. On the basis of the known probability calculation, the state estimation of the football flight trajectory is completed, and the covariance calculation of its error is taken as the input value of the filter control, so that all the variance and mean data are obtained. Finally, the motion state function of football at any time is obtained, and the prediction is completed. The experimental results show that the coincidence degree of the soccer trajectory of this method is 12 mm, and the maximum standard deviation of the impact distance is only 0.0412 m. Therefore, this prediction method can obtain more accurate prediction data.
keywords:RBF network  football penalty kick  parallel filtering  controller  operation track prediction  sensing parameters
查看全文   查看/发表评论   下载pdf阅读器