改进强跟踪EKF算法在MEMS姿态解算中的研究  

Research on improved strong tracking EKF algorithm in MEMS attitude calculation

作　　者：陈志旺[1,2] 姚权允 吕昌昊 郭金华 彭勇 CHEN Zhiwang;YAO Quanyun;LV Changhao;GUO Jinhua;PENG Yong(Hebei Key Laboratory of Industrial Computer Control Engineering,Yanshan University,Qinhuangdao 066004;Key Laboratory of Power Electronics for Energy Conservation and Motor Drive of Hebei Province,Yanshan University,Qinhuangdao 066004;School of Electrical Engineering,Yanshan University,Qinhuangdao 066004)

机构地区：[1]燕山大学工业计算机控制工程河北重点实验室,秦皇岛066004 [2]燕山大学电力电子节能与传动控制河北省重点实验室,秦皇岛066004 [3]燕山大学电气工程学院,秦皇岛066004

出　　处：《高技术通讯》2023年第5期467-478,共12页Chinese High Technology Letters

基　　金：国家自然科学基金(61573305);河北省自然科学基金(F2022203038,F2019203511)资助项目。

摘　　要：本文针对四旋翼姿态解算,提出了一种噪声自适应强跟踪扩展卡尔曼滤波算法(ASTEKF)。当机体从平稳状态向机动状态过渡时,由于量测噪声影响会导致算法估计不准确,因此本文首先证明不同时刻新息序列方差满足正交性原理,正交性原理表明,量测噪声对观测值的准确性影响很大;其次,引入Sage-Husa噪声自适应估计器较准确估计系统量测噪声均值和方差,使观测值更准确;最后,通过满足正交性原理条件公式计算次优渐消因子,将次优渐消因子引入协方差一步预测运算式中,得到强跟踪滤波器。次优渐消因子的引入使得一步预测协方差矩阵增大,即增大强跟踪扩展卡尔曼滤波器增益,使系统增加对观测值权重,得到更准确的状态估计值。离线仿真实验和在线实物实验结果表明了所设计算法的有效性。In this paper,a noise adaptive strong tracking extended Kalman filter(ASTEKF)algorithm is proposed for attitude calculation of quadrotor.When the body transits from stationary state to maneuvering state,the algorithm estimation is not accurate due to the influence of measured noise.Firstly,it is proved that the variance of innovation sequence at different times satisfies the orthogonality principle.The orthogonality principle shows that the measured noise has a great influence on the accuracy of the observed value.Secondly,the mean value and variance of the measured noise are estimated by the Sage-Husa noise adaptive estimator,which leads to getting accurate observed value.Finally,the suboptimal fading factor is calculated by the orthogonality principle,and is used in the one-step prediction of covariance in the strong tracking filter.The suboptimal fading factor leads to an increase in the onestep prediction covariance matrix and the gain of strong tracking extended Kalman filter,that is,increasing of the weight of the observed value makes the state estimation value accurate.The results of offline software simulation and online physical experiment show the effectiveness of the proposed algorithm.

关 键 词：姿态解算 扩展卡尔曼滤波(EKF) 强跟踪滤波器 次优渐消因子 噪声自适应估计器 

分 类 号：V249.1[航空宇航科学与技术—飞行器设计]