Determination of typical patterns from strongly varying signals
Determination of typical patterns from strongly varying signals
复制标题
DOI：
10.1080/10255842.2011.560841
复制
发表时间：
2012-01-01
影响因子：
1.6
通讯作者：
Bergmann, G.
中科院分区：
文献类型：
--
作者：
Bender, A.;Bergmann, G.
当不同的受试者进行步行等活动或重复该活动时，在人体关节中测量到的力会有很大差异。需要“典型”标准化的力-时间模式来测试和改进关节植入物。从机械角度来说，对于它们的耐力而言，最重要的是力的最大值和最小值的大小和时间。它们应该等于单次测量的算术平均值。当评估其他强烈变化的信号时，例如步态分析，也存在类似的问题。计算典型信号 (TS) 的新方法增强了现有的动态时间规整 (DTW) 程序。它允许我们组合任意数量的信号。用于计算 TS 的输入信号序列只有很小的影响。该方法的准确性在可以精确定义典型模式的信号以及不同程度变化的真实关节力上进行了数值测试。
Forces measured in human joints vary considerably when an activity such as walking is carried out by different subjects or when it is repeated. 'Typical' standardised force-time patterns are needed to test and improve joint implants. Mechanically most important for their endurance are the magnitudes and times of force maxima and minima. They should equal the arithmetic means from the single measurements. Similar problems exist when evaluating other strongly varying signals, as in gait analysis. The new method to calculate typical signals (TSs) enhances existing dynamic time warping (DTW) procedures. It allows us to combine any number of signals. The sequence of input signals - used for calculating the TS - has only a minor influence. The accuracy of the method was tested numerically on signals for which the typical patterns could be defined exactly, and also on real joint forces that varied to different extents.