针对无人机的空中对准问题,在有卫星导航的情况下,基于改进OBA(基于优化的对准)对准算法提出了一种统一粗对准和精对准过程的自适应反馈对准算法。在OBA算法后的每一步都进入自适应卡尔曼滤波器进行传感器误差估计和载体系姿态补偿,系统直接输出载体的实时姿态,缩短了对准时间。利用Sage-husa自适应估计器矫正系统的观测噪声协方差矩阵,再构造自适应因子来矫正系统状态量的协方差矩阵,通过KF(卡尔曼滤波器)来估计传感器的误差量,由反馈通道补偿给观测矢量,不断迭代得到姿态角对准结果。设计仿真实验,对卫星导航给出的速度与位置信息存在未知随机误差的情况进行模拟,发现改进后的OBA自适应反馈对准算法有效提高了原OBA算法的对准精度和应用场景,航向角误差由79.45°降低到了-0.032°。文中算法与EKF(扩展卡尔曼滤波)的对比实验,发现文中算法俯仰角与航向角收敛时间较EKF算法缩短了30 s左右,EKF算法收敛后,文中算法误差比EKF算法小2.203°。文中给出算法在量测噪声未知的大失准角初始对准中更有优势,能够应用于大机动飞行器的空中初始对准。

In order to solve the problem of aerial alignment of UAVs under situation of satellite navigation, an adaptive feedback alignment algorithm based on improved OBA(optimization-based alignment) alignment algorithm is proposed to unify the process of coarse alignment and fine alignment. After OBA algorithm, the adaptive Kalman filter is used in each step to estimate the sensor error and compensate the attitude of the carrier. The real-time attitude of the carrier is directly output by the system, and the alignment time is shortened. The Sage-husa adaptive estimator is used to correct the observed noise covariance matrix of the system, and then the adaptive factor is reconstructed to correct the covariance matrix of the system state variables. The sensor error is estimated by KF (Kalman filter), and the observed vector is compensated through the feedback channel and that iterate constantly on the attitude angle alignment results. Simulation experiments are designed and simulate the unknown random errors in the velocity and position information given by satellite navigation, it is found that the improved OBA adaptive feedback alignment algorithm effectively improves the alignment accuracy and application scenarios of the original OBA algorithm, and the heading angle error is reduced from 79.45° to -0.032°. The comparison between the proposed algorithm and EKF (extended Kalman filter) shows that the convergence time of pitch angle and heading angle of the proposed algorithm is about 30 s shorter than that of the EKF algorithm. After the convergence of the EKF algorithm, the error of the proposed algorithm is 2.203° smaller than that of the EKF algorithm. The proposed algorithm has more advantages in the initial alignment for large misalignment angle with unknown measurement noise, and can be applied to the initial alignment of large maneuvering aircraft.