This paper presents a kind of attitude estimation algorithm based on quaternion-vector switching and square-root cubature Kalman filter for autonomous underwater vehicle(AUV).The filter formulation is based on geomagn...This paper presents a kind of attitude estimation algorithm based on quaternion-vector switching and square-root cubature Kalman filter for autonomous underwater vehicle(AUV).The filter formulation is based on geomagnetic field tensor measurement dependent on the attitude and a gyro-based model for attitude propagation. In this algorithm, switching between the quaternion and the three-component vector is done by a couple of the mathematical transformations. Quaternion is chosen as the state variable of attitude in the kinematics equation to time update, while the mean value and covariance of the quaternion are computed by the three-component vector to avoid the normalization constraint of quaternion. The square-root forms enjoy a continuous and improved numerical stability because all the resulting covariance matrices are guaranteed to stay positively semidefinite. The entire square-root cubature attitude estimation algorithm with quaternion-vector switching for the nonlinear equality constraint of quaternion is given. The numerical simulation of simultaneous swing motions in the three directions is performed to compare with the three kinds of filters and the results indicate that the proposed filter provides lower attitude estimation errors than the other two kinds of filters and a good convergence rate.展开更多
由于四元数MUSIC(Multiple Signal Classification)算法计算量较大,本文结合声矢量传感器的四元数导向矢量模型,提出了一种声矢量阵列波达方向估计的四元数最小范数法。首先,将声矢量阵列输出协方差矩阵奇异值分解所得到的(M-N)×M...由于四元数MUSIC(Multiple Signal Classification)算法计算量较大,本文结合声矢量传感器的四元数导向矢量模型,提出了一种声矢量阵列波达方向估计的四元数最小范数法。首先,将声矢量阵列输出协方差矩阵奇异值分解所得到的(M-N)×M维(M为阵元数、N为信源数)噪声子空间依最小范数(Minimum-Norm,MN)准则构建为一个新的四元数域1×M维噪声矢量。接着,提出了简化的谱峰搜索公式,理论分析了四元数最小范数法在搜索计算量上的优势。对提出的算法与Q-MUSIC算法进行了对比。结果显示:该算法至少能节省50%的谱峰搜索量。同时,提出的算法构建的低维噪声矢量与导向矢量间的正交性优于高维噪声子空间与导向矢量间的正交性,在0dB时,其范德蒙范数和谱峰分别为Q-MUSIC算法的1/3和3倍。另外,该算法在减小谱峰搜索量的同时,可以较好地分辨信源波达方向,且其统计特性与四元数MUSIC算法相当。提出的算法不局限于L线阵,也适用于双平行线阵及面阵。展开更多
基金supported by the National Natural Science Foundation of China(1140503561004130+4 种基金60834005)the Natural Science Foundation of Heilongjiang Province of China(F201414)the Postdoctoral Scientific Research Developmental Fund of Heilongjiang Province(LBHQ15034)the Stable Supporting Fund of Acoustic Science and Technology Laboratory(JCKYS2019604SSJS002)the Fundamental Research Funds for the Central Universities。
文摘This paper presents a kind of attitude estimation algorithm based on quaternion-vector switching and square-root cubature Kalman filter for autonomous underwater vehicle(AUV).The filter formulation is based on geomagnetic field tensor measurement dependent on the attitude and a gyro-based model for attitude propagation. In this algorithm, switching between the quaternion and the three-component vector is done by a couple of the mathematical transformations. Quaternion is chosen as the state variable of attitude in the kinematics equation to time update, while the mean value and covariance of the quaternion are computed by the three-component vector to avoid the normalization constraint of quaternion. The square-root forms enjoy a continuous and improved numerical stability because all the resulting covariance matrices are guaranteed to stay positively semidefinite. The entire square-root cubature attitude estimation algorithm with quaternion-vector switching for the nonlinear equality constraint of quaternion is given. The numerical simulation of simultaneous swing motions in the three directions is performed to compare with the three kinds of filters and the results indicate that the proposed filter provides lower attitude estimation errors than the other two kinds of filters and a good convergence rate.
文摘由于四元数MUSIC(Multiple Signal Classification)算法计算量较大,本文结合声矢量传感器的四元数导向矢量模型,提出了一种声矢量阵列波达方向估计的四元数最小范数法。首先,将声矢量阵列输出协方差矩阵奇异值分解所得到的(M-N)×M维(M为阵元数、N为信源数)噪声子空间依最小范数(Minimum-Norm,MN)准则构建为一个新的四元数域1×M维噪声矢量。接着,提出了简化的谱峰搜索公式,理论分析了四元数最小范数法在搜索计算量上的优势。对提出的算法与Q-MUSIC算法进行了对比。结果显示:该算法至少能节省50%的谱峰搜索量。同时,提出的算法构建的低维噪声矢量与导向矢量间的正交性优于高维噪声子空间与导向矢量间的正交性,在0dB时,其范德蒙范数和谱峰分别为Q-MUSIC算法的1/3和3倍。另外,该算法在减小谱峰搜索量的同时,可以较好地分辨信源波达方向,且其统计特性与四元数MUSIC算法相当。提出的算法不局限于L线阵,也适用于双平行线阵及面阵。
