针对无迹卡尔曼滤波(Unscented Kalman Filter,UKF)算法与全球定位系统/惯性导航系统(Global Positioning System/Inertial Navigation System,GPS/INS)组合导航模型不匹配,且鲁棒性不足,难以适应INS 元件的随机性和突变性的问题,提出了一种UKF改进算法。该算法有效结合了混合滤波思想、平方根滤波技术及交互式多模型结构,分别克服了算法与线性/非线性模型不匹配,协方差矩阵非正定以及参数设置难以适应模型不确定性的问题。仿真实验分别考察了新算法在INS平台角初始大误差及加速度计零偏突变两种情况下的表现。实验表明,新算法在估计精度及鲁棒性方面比UKF有较大提高,能够有效校正INS元件产生的随机和突变误差。

For the problem that UKF algorithm does not match with GPS/ INS integrated navigation model, and it is difficult to adapt to randomness and mutations of the INS element for lacking of robustness, a UKF improved algorithm has been proposed. The algorithm combines hybrid filter ideology, square root filtering technology with interacting multi-model structure effectively, overcoming the problem that algorithm does not match with linear/ non-linear model and the algorithm covariance matrix is non-positive, definite difficult to set parameter to adapt the model uncertainty respectively. Simulation experiments were done to investigate the performance of the new algorithm in both cases of the initial INS platform angle error and accelerometer zero bias mutations. The experiments show that the new algorithm has been improved more greatly in the estimation accuracy and robustness than UKF, which can effectively correct the error appearing/ which appears in INS components because of its randomness and mutation.