基于智能手机的健身跑数据采集及演化建模

摘要

针对Brzostowski方法因采集的速度数据精度不高、采用的建模算法搜索能力受限,而导致难以获取高精度健身跑模型的问题,本文提出一种基于智能手机的健身跑速度数据获取及演化建模方法.首先,提出基于智能手机多传感器和中值滤波的健身跑速度采集方法,以滤除由于手机间歇性的姿态变化而产生的三轴加速度信号脉冲噪声,并结合方向传感器对手机三轴加速度中包含的重力分量进行过滤;然后,设计一种健身跑演化建模算法,以增大搜索空间为获取更优的健身跑模型提供支持.实验结果表明,本文提出的方法可以比Brzostowski方法获取更为精确的速度和健身跑运动模型.%Due to the low precision of speed acquisition and the limited search ability of modeling algorithms, it is difficult to obtain a highly precise fitness running model by applying the Brzostowski method. In this paper, we propose a speed acquisition and evolutionarymodeling method for fitness running that is based on the smart phone. First,we introduce a fitness running speed acquisition method based on multiple smartphone sensors and a median filter that can remove the impulse noise of three-axis acceleration signals. This noiseisgenerated by the intermittent gesture changes associated with the use of smartphones. In addition, this method can filter out the gravitational component in three-axis acceleration. Next, we designed an evolutionary modeling algorithm to enlarge the search space to obtain a better fitness running model. The experimental results show that the proposed approach can obtain more accurate speed and fitness running models than the Brzostowski's approach.