摘要
为了提高标准Cubature卡尔曼滤波(CKF)的稳定性和鲁棒性,提出一种改进的多重渐消H∞滤波Cubature卡尔曼滤波算法。首先基于系统状态的可观测性给出多重渐消因子矩阵求解过程,提高滤波算法的稳定性,抑制滤波发散;其次,引入H∞鲁棒思想,构造多重渐消H∞滤波Cubature卡尔曼滤波器;最后,提出采用一种奇异值分解的矩阵分解策略代替标准Cubature卡尔曼滤波中的Cholesky分解,进一步提高算法的数值稳定性。实际GPS/INS组合导航实验表明,改进的多重渐消H∞滤波Cubature卡尔曼滤波算法不仅能有效抑制滤波发散提高算法的稳定性,而且对观测野值具有更高的鲁棒性;提出的新算法与标准CKF算法相比,XYZ三个方向的位置精度分别提高了55.8%,46.6%和39.7%。
In order to improve the stability and robustness of standard cubature kalman filter for INS/GPS integrated navigation nonlinear error model, an improved multiple fading H∞ robust cubature kalman filter algorithm is proposed. First, a multiple fading filtering algorithm is demonstrated based on the observability of the system state. And then a multiple fading H∞ robust cubature kalman filter is improved effectively. In order to get high numerical stability, the singular value decomposition algorithm is used to take the place of Cholesky decomposition in the multiple fading H∞ cubature kalman filter. The actual GPS/INS integrated navigation test indicates that the proposed filter algorithm not can only improve the stability of the algorithm, but also have better robustness to outlier. Compared with standard cubature kalman filter, the navigation precisions of new algorithm are increased by 55.8%, 46.6% and 39.7% in X, Y and Z direction, respectively.
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2013年第4期506-511,共6页
Journal of Chinese Inertial Technology
基金
国家自然科学基金项目(51174206,40904004)
江苏省优势学科建设项目(PAPD,No.SZBF 2011-6-B35)
