研究了一类离散不确定系统中存在等式约束时的最优滤波问题，在均方误差最小的意义下利用卡尔曼滤波给出了最优解。与传统的不确定滤波结果相比，从理论证明了利用更多信息的约束滤波的估计误差协方差的迹更小。-A class of discrete uncertain systems exist in the optimal filter when the equality constraint problem, the minimum mean square error in the sense of Kalman filtering is given optimal solution. Filtering results with the traditional uncertain than proved from the theory of constraints by using more filter the trace of the covariance estimation error smaller.