模型设置的不合理、可能导致卡尔曼滤波器的发散、完成不了预期的最佳滤波功能,这在系统动态噪声较小时尤为明显。本文提出了一种卡尔曼滤波器自我检测发散的方案,并在假设成立时给出修正的方法。使卡尔曼滤波器能尽可能地保持其最佳性、使为了克服发散而牺性的滤波最佳性降至最小。在这个意义上,这种自检的卡尔曼滤波器具有自适应的功能。 The unreasonable model setting may lead to the divergence of the Kalman filter, which can not achieve the optimal filtering function expected, especially when the system dynamic noise is small. In this paper, a self-detecting divergence scheme of Kalman filter is proposed, and a modified method is given when the hypothesis is established. The Kalman filter can be kept as optimal as possible, minimizing the filter optimum sacrificed to overcome the divergence. In this sense, this self-checking Kalman filter has an adaptive function.