Applying Artificial Neural Networks to Improve the Accuracy of Orientation and Speed Parameters for the Carrier Device at the Time of Launch of the Flying Vehicle

This paper presents a method of applying an extended nonlinear Kalman filter to combine information measured from an angular rate gyroscope with magnetometers, accelerometers, and information about the speed of the satellite navigation system to estimate orientation parameters. Based on it, the authors calculate the orientation angle, speed between the coordinate system attached to the carrying device and the navigation coordinate system. In addition, this work also presents a method of applying artificial neural networks to improve the performance of the integrated global positioning system and inertial navigation system during a global positioning system outage. A new model is proposed that directly relates the position, and speed of the inertial navigation system and the speed information of the global positioning system. The combination of an extended nonlinear Kalman filter and multilayer feedforward neural network can estimate the virtual global positioning system speed when the global positioning system signal is not available. In addition, the article also compares the case with only the extended nonlinear Kalman filter without the support of information about the speed of the satellite navigation system. © 2024 The Aeronautical and Astronautical Society of the Republic of China. All rights reserved.