基于递推最小二乘与互补滤波的姿态估计
Attitude estimation based on recursive least square and complementary filtering
摘要点击 151  全文点击 150  投稿时间:2018-05-16  修订日期:2018-08-27
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2018.80359
  2019,36(7):1096-1103
中文关键词  微机电系统  惯性导航  递推最小二乘法  互补滤波器
英文关键词  micro-electromechanical system, inertial navigation, recursive least squares, complementary filters
基金项目  甘肃省基础研究创新群体计划(1606RJIA327), 西部之光青年学者计划(2016XB016), 国家博士后面上基金(2017M613242), 兰州交通大学(201702)优秀平台支持(lzjtu(201702) EP support)资助.
学科分类代码  
作者单位邮编
陈光武 兰州交通大学 自动控制研究所 730070
李少远 上海交通大学 电子信息与电气工程学院 
李文元 兰州交通大学 自动控制研究所 
王迪 兰州交通大学 自动控制研究所 
张琳婧 兰州交通大学 自动控制研究所 
中文摘要
      针对基于微机电系统(Micro-Electromechanical System, MEMS)的惯性导航系统中陀螺噪声较大导致姿态漂移的问题, 本文基于递推最小二乘(Recursive Least Squares ,RLS)与互补滤波器提出一种提高姿态估计精度的方法. 该方法从陀螺去噪算法和姿态解算原理两个方面提高姿态估计精度: 在陀螺去噪方面, 为克服传统递推最小二乘的不足, 提出一种随机加权的递推最小二乘法, 利用随机加权实现对偏差的估计; 在姿态解算方面, 在传统互补滤波器的基础上通过自适应调整比例-积分(Proportional-Integral, PI)参数来调整滤波器的交接频率, 最终得到陀螺积分值的高通滤波和加速度计的低通滤波的叠加. 转台静态和动态实验结果表明,使用本文所提方法后, 有效降低了陀螺噪声, 姿态估计精度明显提升.
英文摘要
      Aiming at the problem of attitude drift caused by gyroscope noise in inertial navigation system based on micro-electromechanical system (MEMS), a method to improve attitude estimation based on recursive least squares (RLS) and complementary filter is proposed. The accuracy of attitude estimation is improved from the aspects of gyro de-noising algorithm and attitude solving principle: in terms of gyro de-noising, in order to overcome the deficiency of traditional recursive least squares, a random weighted recursive least squares method is proposed; in the aspect of attitude calculation, on the basis of the traditional complementary filter, the switching frequency of the filter is adjusted by adaptive proportional-integral (PI) parameter adjustment, and finally the superposition of high pass filtering of gyro integral value and low pass filtering of accelerometer is obtained. The static and dynamic test results of the turntable showed that the proposed method can effectively reduce the noise of gyro and improve the accuracy of attitude estimation.