Full tensor gravity gradient aided navigation based on nearest matching neural network

Advantages of gravity gradient measurement, such as sensitivity to the shallow substance, high accuracy and unsensitivity to the accelerations in the various directions, are with the great significance to the submarine navigation. A distance between the measured full tensor gravity gradients and those predictions from INS and the digital terrain elevation map is defined and a kind of the gravity gradient-aided navigation methods based on nearest matching neural network is proposed in this paper. In the novel navigation systems, the measured full tensor gravity gradients is as inputs of nearest matching neural network, the full tensor gravity gradients evaluations is as weights between the input layer and the middle layer of nearest matching neural network, the output function is defined and the variable interested domain matching strategy is adopted to correct the INS errors. Simulation results show that an ideal matching probability can be got.