A New Geomagnetic Vector Navigation Method Based on a Two-Stage Neural Network

The traditional geomagnetic matching navigation method is based on the correlation criteria operations between measurement sequences and a map. However, when gradient of field small, there are multiple similar data in database to value, which means correlation-based fails. Based idea pattern recognition, this paper constructs two-stage neural network by cascading probabilistic non-fully connected to, respectively, classify vectors their feature information two steps: “coarse screening” “fine screening”. effectiveness accuracy vector algorithm verified through simulation experiments. In simulation, it that average 5 nT/km, fails, while positioning proposed 40.17 m, success rate also reaches 98.13%. Further, flight experiments, under an 11 error 39.01 99.42%.