包含乘性噪声自适应修正的非合作目标相对导航算法

无人机（UAV）态势感知的任务是利用机载传感器对未知环境进行目标识别和引导，针对无人机与非合作目标间中远距离的相对导航问题，提出了一种基于角度和距离量测的相对状态估计算法。在现有滤波算法的基础上，为了提高精度和稳定性，本文利用了列文伯格-马夸尔特（LM）优化的思想对迭代卡尔曼滤波（IEKF）算法进行改进，提出了一种LM-IEKF算法，并推导该算法在迭代过程中的状态更新方程及协方差阵的递推公式。在此基础上，考虑到距离传感器由于信号相关特性而引入的乘性噪声，现有的加性噪声模型难以适应，因此，进一步提出了基于量测噪声自适应修正的Modified LM-IEKF方法，通过在线实时更新噪声阵提高滤波的精度，并设置渐消记忆指数平滑估计结果。算法验证结果表明，与现有的EKF、IEKF算法相比，在仅含加性噪声的情况下，LM-IEKF算法具有更好的性能；在包含乘性噪声的情况下，Modified LM-IEKF可以有效地估计量测噪声，与目前广泛使用的EKF算法相比，在综合相对位置和相对速度精度上分别提高了10%和23%。

Relative navigation algorithm for non-cooperative target with adaptive modification of multiplicative noise

The task of Unmanned Aerial Vehicle (UAV) situational awareness is to use airborne sensors to identify and guide targets in unknown environments. To solve the relative navigation problem between UAV and non-cooperative target at medium and long distances, a relative state estimation algorithm based on angle and distance measurements is proposed. On the basis of existing filtering algorithms, in order to improve the accuracy and stability, this paper improves the Iterative Extended Kalman Filter (IEKF) algorithm by using the Levenberg-Marquardt (LM) optimization. This paper proposes a Levenberg-Marquardt Iterative Extended Kalman Filter (LM-IEKF) algorithm, deducing the state updating equation and the recursive formula of covariance matrix in the iteration process of the LM-IEKF algorithm. The multiplicative noise introduced by the distance sensor due to the signal correlation characteristics makes the existing additive noise model difficult to adapt. Therefore, a Modified LM-IEKF method based on adaptive correction of measurement noise statistics is proposed. The filtering accuracy is improved by updating the noise matrix online in real time. Meanwhile, the results of fading memory index smoothing estimation are set up. Verification results of the algorithm show that compared with the existing EKF (Extended Kalman Filter) and IEKF algorithms, the LM-IEKF algorithm has better performance in additive noise only situation. In the case of containing multiplicative noise, Modified LM-IEKF can effectively estimate the measurement noise. Compared with the widely used EKF algorithm, the accuracy of integrated relative position and relative velocity improved by 10% and 23%.