神经网络辅助的组合导航算法研究  

Research on neural network-assisted integrated navigation algorithm

作　　者：白相文 杨建华[1] 杨志强[1] BAI Xiangwen;YANG Jianhua;YANG zhiqiang(School of Geology Engineering and Geomatics,Chang’an University,Xi’an 710054,China)

机构地区：[1]长安大学地质工程与测绘学院

出　　处：《导航定位学报》2020年第1期93-98,共6页Journal of Navigation and Positioning

摘　　要：针对在高动态、强干扰的环境下,单一导航源无法满足导航要求,组合导航技术应运而生,其中捷联惯导系统(SINS)及全球定位系统(GPS)组合导航应用最为广泛。当卫星导航信号较弱甚至失锁时,单一的惯性导航误差逐渐积累,无法满足导航精度要求,提出1种埃尔曼(Elman)神经网络辅助的组合导航联邦滤波算法,基于Elman神经网络内部反馈特性和联邦滤波的并行处理能力,在GPS信号失锁前对神经网络实施训练,当卫星信号失锁时启用神经网络实时预测与解算。最后通过MATLAB仿真实验证明了该算法可以在卫星信号失锁时有效抑制惯导误差的累积,具有一定的参考价值。In the environment of high dynamics and strong interference,the satellite navigation signal is weak or even unlocked,and a single inertial navigation error accumulatescan not meet the navigation accuracy requirements.An Elman neural network-assisted integrated navigation federated filtering algorithm is proposed,Based on the feedback characteristics of the Elman netural network and the parallel processing capability of the federated filte perform,We first train the neural network before the GPS signal is lost.When the satellite signal is out of lock,the neural network is used to implement prediction and solution.The MATLAB simulation experiment proves that the algorithm can effectively suppress the accumulation of inertial navigation errors when the satellite signal is out of lock,which has some reference value.

关 键 词：神经网络 联邦滤波 组合导航 数据融合 失锁 

分 类 号：P228[天文地球—大地测量学与测量工程]