Comparing the Kalman filter with a Monte Carlo-based artificial neural network in the INS/GPS vector gravimetric system

Rigorous physical and mathematical analysis has been intensively developed to obtain the gravity disturbance vector from the inertial navigation system and the global positioning system. However, the combination of the observation noise and the systematic INS errors make it very challenging to accurately and efficiently describe the dynamics of the system with rigorous equations. Thus, the accuracy of the gravity disturbance estimates, especially in the horizontal components, is limited by the insufficient error models. To overcome the difficulty of directly modeling the systematic errors with exact mathematical equations, a Monte Carlo based artificial neural network is successfully applied in the moving base gravimetric system. The computation results show significant improvement in the precision of all components of the gravity disturbance estimates.