用于MEMS陀螺阵列信号处理的OBE平滑算法（英文）

OBE smoother for signal processing of MEMS gyroscope array

为提高MEMS陀螺的精度,提出了一种基于最优定界椭球(OBE)的平滑算法,并将其用于陀螺阵列信号的处理。首先,利用多个相同型号的MEMS陀螺构成阵列,测量同一角速率信号,并建立数据融合模型。对于融合问题而言,噪声统计特性的不确定会导致传统融合方法精度下降。为解决该问题,引入仅要求噪声未知但有界的集员估计理论,结合RTS平滑思想,提出一种新的平滑算法作为融合方法,它由前向滤波和反向平滑两个过程构成:前者采用集员估计理论中的OBE滤波估计角速率,后者则逆序执行OBE算法进一步提高估计精度。实验表明:该方法能够将陀螺的静态漂移由0.5130(°)/s降低到0.1368(°)/s;动态条件下,在有效跟踪载体角度变化的同时,将漂移由0.5343(°)/s降低到0.1704(°)/s,显著提高了陀螺的使用精度。

To improve the accuracy of the MEMS gyroscopes, a novel smoother based on the optimal bounding ellipsoid（OBE） algorithms is proposed for signal processing of the gyroscope array. Firstly, several gyroscopes measuring the same angular rate signal are used to construct an array, and then a data fusion model is established. For traditional fusion methods, the uncertainty of noises statistical characters may lead to accuracy degeneration. To solve this problem, the set-membership（SM） methods, in which errors are assumed only to be bounded, can be used as a substitute. The proposed SM method is a Rauch-Tung-Striebel（RTS）-type smoother, which is constructed by a forward pass and a backward one. The forward pass estimates the angular rate with the OBE filter, while the backward pass updates the estimates by using the OBE algorithm in the backward direction. Experiment results indicate that the method can significantly improve the accuracy of MEMS gyroscopes, with the static drift of the estimated rate signal being reduced to 0.1368（°）/s from 0.5130（°）/s. Under dynamic condition, the drift can be reduced to 0.1704（°）/s from 0.5343（°）/s.