基于Kalman滤波的一阶微分参量估计方法的研究

基于迭代Kalman滤波算法，提出了一种微分参量的估计方法，并将其应用于腕力传感器的一阶力微分信号的提取处理中．通过对观测信号建立AR模型和递推的使用Kalman滤波算法，有效地抑制了噪声强度从而提高了微分参量的计算精度，克服了直接计算法误差较大的缺点，同时还避免了因加装速度传感器而对原腕力传感器动态性能造成的影响．文中讨论了在均匀分布的背景噪声下如何估计原始信号的一阶微分参量的问题，并给出了仿真结果．试验表明该方法具有良好的计算精度和较强的收敛性．

Research on one-order differential coefficient estimation based on Kalman filtering

Based on iteration of the Kalman filter, a method is proposed to estimate the differential coefficient and to be applied to get the one order differential coefficient from a wrist transducer. Through constructing an AR model of measured signal and recursive employment of the Kalman filter, the strength of background noise is effectively controlled, so the computation accuracy is attained and the serious error of the direct computation method is overcome. Meanwhile, as will eliminate the influence on dynamic performance of wrist transducer due to surplus apparatus. The problem of how to extract differential information from the signal contaminated by uniform distribution noise is discussed and experimental results are presented. The experiment shows that a high performance and robustness can be achieved by this method.