基于Bagging模型的惯导系统误差抑制方法

发布时间：2018-03-05 11:01

本文选题：GPS失锁　切入点：惯性导航系统　出处：《中国惯性技术学报》2017年01期 　论文类型：期刊论文

【摘要】：由惯性导航系统(SINS)和卫星导航系统(GPS)构成的组合导航系统一直是陆用车辆的主要导航设备。当GPS失锁时,SINS的定位误差将随着时间不受控制的迅速增长。为了提高惯导系统的定位精度,相比较于单一的神经网络,集成学习算法中的Bagging模型能够深度学习惯导误差之间的内在关系,进一步提高导航性能。在智能算法和组合导航系统的框架下提出了惯导系统的误差抑制方案,即在GPS存在时训练组合导航系统数据,当GPS失锁时预测惯导系统位置增量。试验结果表明,该方案能够在GPS丢失时抑制惯导系统定位误差发散,相比较于BP算法,Bagging模型的定位精度在5 min时提高了约49%,15 min时提高了约41%。
[Abstract]:The integrated navigation system composed of inertial navigation system (ins) and satellite navigation system (GPS) has always been the main navigation equipment for land vehicles. The positioning error of sins will increase rapidly with time when GPS is lost. Positioning accuracy of Inertial Navigation system, Compared with a single neural network, the Bagging model in the ensemble learning algorithm can deeply learn the inherent relationship between the habit guide errors, Under the framework of intelligent algorithm and integrated navigation system, the error suppression scheme of inertial navigation system is proposed, that is, training the integrated navigation system data in the presence of GPS. The experimental results show that the proposed scheme can suppress the divergence of ins positioning errors when GPS is lost, and the accuracy of this scheme can be improved by about 41% when the BP algorithm bagging model is about 5 min higher than that of BP algorithm at 5 min.
【作者单位】：北京航空航天大学仪器科学与光电工程学院;北京航天控制仪器研究所;陆军装备部;国网江苏省电力公司检修分公司;
【基金】：国家安全重大基础研究资助项目(613186) 中央高校基本科研业务费专项资金资助项目(YWF-10-01-B30)
【分类号】：U495;U463.67

【相似文献】