Gravity gradient aided position approach based on EKF and NN

A gravity gradient aided position approach based on Extended Kalman Filter (EKF) is proposed in this paper. The characteristics of gravity gradient measurement have great significance to the underwater position. The underwater carrier's current position obtained by EKF, whose input is the difference of the gravity gradient values measured by gradiometer and the predicted gravity gradient values got from the reference map, to correct the inertial navigation system's accumulated error. Owing to the sensitivity of gravity gradient to terrain, the gravity gradient reference map can be prepared from the local terrain elevation data. To avoid the divergence of Kalman filter, gravity gradient linearization techniques on neural network (NN) is used. The simulation results show that gravity gradient linearization techniques on neural network is more effective than the Nine-point Fit (NPF) linearization technique in gravity gradient aided position approach based on EKF.