采用卡尔曼滤波方法,针对带有观测时滞的线性离散系统,研究了输入白噪声最优估计器的设计.通过对观测序列进行重新组织,使之成为无时滞的观测,并进一步给出重组新息序列.由Hilbert空间上的正交投影定理,通过求解与原系统同维的两个Riccati方程实现递推计算.该方法能避免状态扩维,有效地减轻了计算负担.最后通过仿真实例说明该方法的有效性. The Kalman filter method is used to design the optimal input white noise estimator for a linear discrete system with observation time delay. By reorganizing the observed sequence into a time-delay-free observer, Reorganization interest sequence.Orthogonal projection theorem in Hilbert space is used to realize recursive computation by solving two Riccati equations which are in the same dimension with the original system.It can avoid the state expansion and reduce the computational burden effectively.At last, The example shows the effectiveness of this method.