常规卡尔曼滤波器有一大缺点，它要求精确的模型和噪声统计，但实际问题中，大多数情况上述要求不能满足。本文给出了考虑滤波器初始条件不精确性的Ｋａｌｍａｎ滤波器表达式。利用这种Ｋａｌｍａｎ滤波方法进行故障检测，降低了误报警率，提高了鲁棒性。由于导弹姿态控制系统模型的精确性有限，常规Ｋａｌｍａｎ滤波方法因其鲁棒性差很难检测故障，这种方法能有效检测故障。仿真算例表明此方法非常有效。 The conventional Kalman filter has a major drawback, which requires accurate model and noise statistics, but in most cases the above requirements can not be met. In this paper, the Kalman filter expression considering the inaccuracy of the initial conditions of the filter is given. Using this Kalman filtering method to detect faults, the false alarm rate is reduced and the robustness is improved. Due to the limited accuracy of the missile attitude control system model, the conventional Kalman filtering method is difficult to detect faults because of its poor robustness. This method can effectively detect faults. The simulation example shows that this method is very effective.