Geneau Daniel, Cormier Patrick, Tsai Ming-Chang, Agar-Newman Dana, Lenetsky Seth, Klimstra Marc
Canadian Sport Institute Pacific, Victoria, BC V9E 2C5, Canada.
School of Exercise Science, Physical and Health Education, University of Victoria, Victoria, BC V8P 5C2, Canada.
Sensors (Basel). 2024 Nov 29;24(23):7632. doi: 10.3390/s24237632.
Accurate linear sprint modelling is essential for evaluating athletes' performance, particularly in terms of force, power, and velocity capabilities. Radar sensors have emerged as a critical tool in capturing precise velocity data, which is fundamental for generating reliable force-velocity (FV) profiles. This study focuses on the fitting of radar sensor data to various sprint modelling techniques to enhance the accuracy of these profiles. Forty-seven university-level athletes (M = 23, F = 24; 1.75 ± 0.1 m; 79.55 ± 12.64 kg) participated in two 40 m sprint trials, with radar sensors collecting detailed velocity measurements. This study evaluated five different modelling approaches, including three established methods, a third-degree polynomial, and a sigmoid function, assessing their goodness-of-fit through the root mean square error (RMSE) and coefficient of determination (r). Additionally, FV metrics (, , , , and ) were calculated and compared using ANOVA.
Significant differences ( < 0.001) were identified across the models in terms of goodness-of-fit and most FV metrics, with the sigmoid and polynomial functions demonstrating superior fit to the radar-collected velocity data.
The results suggest that radar sensors, combined with appropriate modelling techniques, can significantly improve the accuracy of sprint performance analysis, offering valuable insights for both researchers and coaches. Care should be taken when comparing results across studies employing different modelling approaches, as variations in model fitting can impact the derived metrics.
精确的直线冲刺建模对于评估运动员的表现至关重要，特别是在力量、功率和速度能力方面。雷达传感器已成为获取精确速度数据的关键工具，而精确速度数据是生成可靠的力-速度（FV）曲线的基础。本研究聚焦于将雷达传感器数据与各种冲刺建模技术进行拟合，以提高这些曲线的准确性。47名大学水平的运动员（男性 = 23名，女性 = 24名；身高1.75 ± 0.1米；体重79.55 ± 12.64千克）参加了两次40米冲刺试验，雷达传感器收集了详细的速度测量数据。本研究评估了五种不同的建模方法，包括三种既定方法、一个三次多项式和一个Sigmoid函数，通过均方根误差（RMSE）和决定系数（r）评估它们的拟合优度。此外，使用方差分析计算并比较了FV指标（，，，，和）。
在拟合优度和大多数FV指标方面，各模型之间存在显著差异（< 0.001），Sigmoid函数和多项式函数对雷达收集的速度数据显示出更好的拟合效果。
结果表明，雷达传感器与适当的建模技术相结合，可以显著提高冲刺表现分析的准确性，为研究人员和教练提供有价值的见解。在比较采用不同建模方法的研究结果时应谨慎，因为模型拟合的差异可能会影响得出的指标。
