基于高斯混合模型的惯导/计程仪组合导航方法

SINS/EML navigation method based on Gaussian mixtures unscented Kalman filter

针对长航时舰船航行过程中电磁计程仪误差变化较大,同时存在未知测量噪声,无法满足船用捷联惯导/电磁计程仪组合导航系统对计程仪要求的问题,提出了一种用于非线性非高斯系统状态估计的滤波方法。以无迹卡尔曼滤波为组合导航系统基本算法,测量噪声密度分布中引入高斯混合模型,提出了捷联惯导/电磁计程仪组合导航的高斯混合模型无迹卡尔曼滤波算法,达到实时准确估计并补偿惯性导航系统误差的目的。航行试验验证了基于高斯混合模型组合导航方法的可行性,使得捷联惯导/电磁计程仪组合导航系统的最大定位误差由水平阻尼的1213 m减小到392 m,且比传统无迹卡尔曼滤波方法进一步消除了计程仪误差的影响,定位精度提高了15%。

In view of the large variation of Electromagnetic Log(EML) error in ship navigation and the presence of unknown measurement noise, a filtering method for the state estimation of nonlinear and non-Gaussian systems is studied for the shipboard strapdown inertial navigation system(SINS)/EML integrated navigation system. Taking the unscented Kalman filter as the basic algorithm of the integrated navigation system, and introducing the Gaussian mixture(GM) model into the measurement of noise density distribution, a Bayesian unscented Kalman filtering algorithm of SINS/EML integrated navigation is proposed to accurately estimate and compensate the errors of the combined navigation system in real time. Voyage test proves the effectiveness of the proposed method, and shows that the maximum positioning error of the SINS/EML integrated navigation system is reduced from 1213 m to 392 m. The experimental results also show that the proposed method further eliminates the influence of the Log error, and the positioning accuracy is improved by 15% compared with that of the traditional unscented Kalman filter method.