Applying Neural Networks to Improve the Accuracy of the Directional Cosine Matrix Between the Gimbal Coordinate System and Geographic Coordinate System

To control a flying vehicle in a gimbal inertial navigation system, it is necessary to accurately determine the orientation angles of the gimbal coordinate system relative to the navigation coordinate system as well as the orientation angle of the associated coordinate system in comparison to the navigation coordinate system (geographic coordinate system). The authors propose a solution to place a micromechanical gyroscope measuring angular speed, an accelerometer, and a magnetometer on the carrier device, combining information about the speed of the satellite navigation system to estimate the navigation parameters of the carrier device through an extended nonlinear Kalman filter. The paper focuses on improving the speed accuracy of the carrier device when losing GPS at the time of launching the flying vehicle by applying a multilayer artificial neural network to improve the accuracy when losing GPS. Based on accurately determining the speed of the carrying device, the direction Cosine matrix between the gimbal coordinate system mounted on the automatic flight device and the geographical coordinate system will be accurately determined by the method of coordinating the velocity vector when the carrying device moves at a low speed. This issue is still new and little has been published.