CAJ | 학술논문

研究带自回归滑动平均(ARMA)有色观测噪声的多传感器广义离散随机线性系统,根据Kalman滤波方法和白噪声估计理论,在线性最小方差信息融合准则下,应用奇异值分解和增广状态空间模型,为了提高融合器的精度,提出了按矩阵加权降阶稳态广义Kalman融合器,可统一处理稳态滤波、平滑和预报问题,可减少计算负担和改善局部估计精度.并提出最优加权系数的局部估计误差方差和协方差阵的计算公式.用一个Monte Carlo数值仿真实例说明了所提方法的有效性.
연구대자회귀활동평균(ARMA)유색관측조성적다전감기엄의리산수궤선성계통,근거Kalman려파방법화백조성고계이론,재선성최소방차신식융합준칙하,응용기이치분해화증엄상태공간모형,위료제고융합기적정도,제출료안구진가권강계은태엄의Kalman융합기,가통일처리은태려파、평활화예보문제,가감소계산부담화개선국부고계정도.병제출최우가권계수적국부고계오차방차화협방차진적계산공식.용일개Monte Carlo수치방진실례설명료소제방법적유효성.
The multi-sensor generalized discrete stochastic linear system with autoregressive moving average (ARMA)colored observation noises are studied.Based on Kalman filtering method and white noises estimation theory,a reduced order steady-state generalized Kalman fuser weighted by matrices is proposed under the linear minimum variance information fusion criterion by using the singular value decomposition and augmented state space model.It can handle the fused filtering,smoothing and prediction problems in a unified framework,and can reduce the computational burden and improve the accuracy of local estimation.The formulas for computing variance and covariance matrices among local estimation errors are presented and applied to obtain the optimal weighting coefficient.A Monte Carlo numerical simulation example shows the effectiveness of the proposed method.