一种改进的自适应指数加权衰减记忆滤波算法  

An Adaptive Filter Algorithm of Exponentially-weighted Multiple Attenuated Memory

作　　者：田宠[1] 王兴亮[1] 卢艳娥[1] 

机构地区：[1]空军工程大学电讯工程学院,西安710077

出　　处：《武汉理工大学学报》2011年第11期147-152,共6页Journal of Wuhan University of Technology

基　　金：国家自然科学基金(61071014);学院科研创新基金(DYCX1002)

摘　　要：对卡尔曼滤波在GPS定位应用中估值发散的抑制进行了研究,首先解决了由于各个测量噪声之间相关导致噪声协方差阵无法满足对角矩阵的问题;接着对基于多重衰减因子矩阵的指数加权的衰减自适应算法的一步预测均方误差进行了UD分解,将分解矩阵应用于卡尔曼滤波过程更新,并给出了多重衰减记忆因子矩阵的求解过程。实验结果表明,这种算法增强了数值稳定性,提高了定位精度,具有抑制滤波发散的作用。This thesis firstly research the phenomenon of valuation instability in which Kalman filtering process in GPS positioning.The method of orthogonal matrix decomposition is used to solve the failure of noise covariance matrix variance in meeting diagonal matrix,which is led by the correlation between various measurement noises.Then by the UD matrix decomposition of the one step forecast mean square error in which adaptive algorithm of multiple exponential weighting fading factors matrix,the decomposition matrix is applied in updating processing of the Kalman filtering,as well,the solving process of multiple fading factors matrix is provided.The simulation result demonstrates that the algorithm enhances valuation stability,improves positioning accuracy and has the function of preventing filter divergence thus verifies the effectiveness of this algorithm.

关 键 词：UD矩阵分解 噪声方差阵 衰减记忆因子矩阵 定位精度 

分 类 号：TN967.1[电子电信—信号与信息处理]