基于全连接神经网络的GNSS/INS组合定位方法
王尔申 , 刘依凡 , 于腾丽 , 杨健 , 刘达 , 张树宁 , 易静怡 , 李昕
沈阳航空航天大学学报 ›› 2024, Vol. 41 ›› Issue (05) : 54 -61.
基于全连接神经网络的GNSS/INS组合定位方法
GNSS/INS integrated positioning method based on fully connected neural network
作为一种比单一导航系统精度更高、稳健性更好的导航方案,全球导航卫星系统/惯性导航系统(global navigation satellite system/inertial navigation system,GNSS/INS)组合导航,在各种运动载体中得到广泛应用。针对环境遮挡或电磁干扰导致卫星导航信号中断从而降低组合定位精度的问题,提出一种基于全连接神经网络(fully connected neural network,FCNN)的GNSS/INS组合导航定位方法,该方法由训练模块和预测模块构成。在GNSS信号正常的情况下,该方法利用INS解算的位置和速度信息与组合导航输出的位置和速度信息对FCNN模型进行训练;当GNSS信号中断或失效时,利用前期训练的FCNN模型预测输出导航解。采用实测数据对所提方法进行了验证,结果表明,提出的基于FCNN的GNSS/INS组合导航方法可有效地抑制单一惯导定位误差发散的情况,提高GNSS信号中断时GNSS/INS组合定位结果的精度和可用性。
As a navigation solution with higher accuracy and better robustness compared to single navigation systems,GNSS/INS integrated navigation has been widely applied in various carriers.Aiming at the problem of the satellite navigation signal interruption caused by environmental obstruction or electromagnetic interference to reduce the accuracy of integrated positioning,GNSS/INS integrated navigation positioning method by a fully connected neural network (FCNN)was proposed.This method consisted of training and prediction modules.Under normal GNSS signal conditions,the me-thod utilized the position and velocity information calculated by INS and the position and velocity information output by the integrated navigation to train the FCNN model.When the GNSS signals were interrupted or fail,the pre-trained FCNN model was used to predict the navigation solutions.Experimental data was employed to validate the proposed method.The results indicate that the GNSS/INS based on FCNN integrated navigation method proposed in this study effectively suppresses the divergence of single INS positioning errors,thereby improving the accuracy and availability of GNSS/INS integra-ted positioning results when the GNSS signals are interrupted.
| [1] |
|
| [2] |
|
| [3] |
王尔申,朱骏,徐嵩, |
| [4] |
王尔申,孙彩苗,黄煜峰, |
| [5] |
|
| [6] |
高为广,陈谷仓.结合自适应滤波和神经网络的GNSS/INS抗差组合导航算法[J].武汉大学学报(信息科学版),2014,39(11):1323-1328. |
| [7] |
徐晓苏,周峰,张涛, |
| [8] |
|
| [9] |
陈光武,李文元,于月, |
| [10] |
陈光武,程鉴皓,杨菊花, |
| [11] |
|
| [12] |
|
| [13] |
|
| [14] |
张跃发.神经网络辅助的GNSS/INS车载组合导航算法研究[D].西安:长安大学,2023. |
| [15] |
|
| [16] |
|
| [17] |
|
国家自然科学基金(62173237)
嵩山实验室预研项目(YYJC062022017)
辽宁省科技厅应用基础研究计划项目(2022020502-JH2/1013)
辽宁省科技厅应用基础研究计划项目(2022JH2/101300150)
沈阳市科技局项目(22-322-3-34)
辽宁省教育厅重点攻关项目(LJKZZ20220031)
/
| 〈 |
|
〉 |