基于卡尔曼滤波的无人机姿态测量研究∗

摘要

Subject to the limitations of computer storage and processing parameters of each byte, which makes the precision of calculation is also subject to certain restrictions, rounding errors inevitably exist in the system, the errors of Kalman filter state estimation can be corrected by the update process of measured ,but the error covariance matrix P can not be corrected, along with the iteration number increased, the covariance matrix P will become distortion . It is diagonal symmetry will broken. In order to improve the attitude meas-urement accuracy of UAV, thus design the attitude measurement system composed of gyroscope accelerome-ter and magnetometer, using the method of UD decomposition EKF to data fusion to estimate Quaternion and verified by the experiment. The experimental results show that the UD decomposition EKF method can accu-rately calculate the attitude angle in low dynamic environment, and can effectively overcome the influence of rounding errors.%文章针对无人机姿态测量中受制于计算机存储和处理每个参数字节的限制，使得计算精度也同样受到一定限制，舍入误差不可避免的存在于系统之中，对于卡尔曼滤波状态估计的误差可以由测量更新过程来修正，然而对误差协方差矩阵P却没有相应的修正，随着迭代次数不断增加，将导致P矩阵失真，打破其对角对称性。为提高无人机姿态测量的准确性，由此设计由陀螺仪、加速度计和磁强计组成的姿态测量系统，采用UD分解扩展卡尔曼滤波方法，实现对四元数的估计，并通过实验进行分析验证，实验结果表明该UD分解扩展卡尔曼滤波方法能够在低动态环境下准确解算出姿态角度，并能有效克服舍入误差的影响。