论文部分内容阅读
针对SINS/GPS组合导航中量测噪声统计特性不准确引起卡尔曼滤波精度下降的问题,提出基于变分贝叶斯自适应无迹卡尔曼滤波(VB-UKF)的非线性融合方法.分析了线性的变分贝叶斯自适应卡尔曼滤波(VB-KF)算法的原理与性能,针对其仅适用于线性系统的问题,将VB-KF与UKF结合导出了非线性的VB-UKF算法.该算法可对系统状态和时变的量测噪声方差进行同步非线性估计,且与传统的UKF算法具有统一的形式.导航仿真结果表明:VB-UKF对于突变或慢变的量测噪声方差均能实时跟踪,较常规UKF算法可有效降低噪声统计特性不准确给系统造成的不利影响,提高定位精度.
Aiming at the problem that the accuracy of Kalman filter is not accurate due to the inaccuracy of the measurement noise in SINS / GPS integrated navigation, a nonlinear fusion method based on Variational Bayesian adaptive unscented Kalman filter (VB-UKF) is proposed. The principle and performance of the linear variant Bayesian adaptive Kalman filter (VB-KF) algorithm are studied. The VB-UKF algorithm derived from VB-KF and UKF is derived for the problem that it is only applicable to linear systems. The proposed algorithm can synchronously and nonlinearly estimate the state noise and the time-varying measurement noise variance, and has a uniform form with the traditional UKF algorithm. The simulation results show that the VB-UKF measures the noise variance both abruptly or slowly Can track in real time. Compared with the conventional UKF algorithm, it can effectively reduce the unfavorable influence caused by inaccurate noise statistics and improve the positioning accuracy.