A geomagnetic model error prediction method based on artificial neural networks is proposed in response to the difficulty of improving geomagnetic navigation precision because of the slow update and low accuracy of the geomagnetic field model.The mapping relationship is combined with the magnetic field measurement data of on-orbit satellites to estimate and predict the geomagnetic field model error by establis-hing the relationship among geomagnetic field intensity vector elements,geographical features,and time in-formation.m addition,a geomagnetic navigation method combining neural network with filter is proposed.In order to verify the accuracy and effectiveness of this method,simulation verification is conducted by using satellite measured information.The results show that compared with the advanced filter method in recent years'literatures,the position and velocity accuracy can be improved from 4.15 km,4.38 m/s to 1.34 km,1.47 m/s that significantly improves the precision and effect of geomagnetic navigation.
关键词
地磁导航/地磁模型误差预测/BP神经网络/地磁滤波
Key words
Geomagnetic navigation/Geomagnetic model error predicts/BP neural network/Geomagnetic filtering
维普
万方数据