基于冗余测量信息的阵列式IMU设计

Array IMU design based on redundant measurement information

针对单个微机电系统(MEMS)陀螺仪精度低、可靠性差的问题,基于冗余测量信息设计了阵列惯性测量单元(IMU)硬件平台,并利用迭代重加权最小二乘法(IRLS)对扩展卡尔曼滤波(EKF)融合算法进行了改进。首先,根据惯性器件的误差特性及冗余测量可靠性确定了IMU数量,并设计了同轴反向、平行正交的阵列IMU硬件平台;然后,结合受力分析,使用IRLS对数据进行降维融合,解决了使用EKF融合算法时,由于矩阵维数高导致求逆运算复杂和受离群值影响导致融合精度下降的问题;最后,通过实验和仿真证明:阵列IMU的随机误差比单个IMU降低了3~5倍;当出现离群测量值时,阵列IMU仍能保证较高的精度,具有较高的可靠性。

Aiming at the problem of low precision and poor reliability of a single MEMS gyroscope,an array inertial measurement unit(IMU)hardware platform is designed based on redundant measurement information,and iteratively reweighted least square(IRLS)is used to improve the extended Kalman filtering(EKF)fusion algorithm.Firstly,the number of IMU is determined according to the error characteristics of the inertial device and the reliability of redundant measurement,and the coaxial reverse,parallel orthogonal array IMU hardware platform are designed.Then,combined with the force analysis,IRLS is used to reduce the dimensionality of the data and fuse,solve the problem of complex inversion calculations due to high matrix dimension and fusion precision reducing caused by outliers when using the EKF fusion algorithm.Finally,experiments and simulations prove that the random error of the array IMU is 3~5 times lower than that of a single IMU.When outlier measurement values appear,the array IMU can still guarantee higher precision and higher reliability.