分段式 AEKF 高速弹丸组合导航算法

In the integrated navigation system of high speed projectile, the statistical characteristics of measurement noise cannot be accurately obtained because the projectile is affected by the complex environment in the air, its high-speed spin, vibration and other factors, which leads to the decline of the precision of the integrated navigation system. Therefore, we propose a segmented AEKF high-speed projectile integrated navigation method based on convolutional neural networks (CNN). Under the framework of GNSS/SINS loose combination, we trained the CNN model to identify the flight phase of the projectile in real time. The measurement noise covariance matrix of AEKF is adjusted adaptively by associating each flight stage with the noise estimator parameters of AEKF algorithm, and the precision of integrated navigation is improved. The proposed method is tested and compared with the general AEKF projectile integrated navigation method on the same data set. The test results show that the proposed method reduces the root mean square errors of speed and position by an average of 54% and 43%, which has good reference and application value.