基于卷积神经网络的冗余高精度加速度计导航系统研究  

A Study on Redundant High-Precision Accelerometer Navigation System Based on Convolutional Neural Networks

作　　者：贺钦源 于化鹏 梁大雷 杨小卓 HE Qinyuan;YU Huapeng;LIANG Dalei;YANG Xiaozhuo(National Innovation Institute of Defense Technology,Academy of Military Sciences,Beijing 100071,China;University of Electronic Science and Technology of China,School of Automation Engineering,Chengdu 611731,China)

机构地区：[1]军事科学院国防科技创新研究院,北京100071 [2]电子科技大学自动化学院,成都611731

出　　处：《智能安全》2024年第3期66-74,共9页Artificial Intelligence Security

摘　　要：惯性导航系统(Inertial Navigation System,INS)是为各类运载体平台在GPS拒止环境下提供自主导航的关键技术。然而,现有的INS存在误差累积问题,影响其长期导航精度。为增强INS的准确性,本文通过引入高精度加速度计并利用卷积神经网络(Convolutional Neural Networks,CNN)提出一种新型基于卷积神经网络的冗余高精度加速度计导航系统,该系统优化了误差预测和反演校正能力,显著提高了导航精度。实验结果证明,该系统在减少误差累积和提高长期导航稳定性方面优于传统INS方法。The Inertial Navigation System(INS)is a pivotal technology for autonomous navigation in GPS-denied environments.However,existing INS systems are confronted with challenges of error accumulation,which affects long-term navigational accuracy.To enhance the accuracy of INS,by introducing high-precision accelerometers and using Convolutional Neural Network(CNN),this paper proposes a new type of CNN-based redundant high-precision accelerometer navigation system,which optimizes the error prediction and inversion correction capabilities,and significantly improves the navigation accuracy.Experimental results prove that this system is superior to the traditional INS methods in reducing error accumulation and improving the long-term navigation stability.

关 键 词：惯性导航 冗余结构 高精度加速度计 卷积神经网络 

分 类 号：U666.12[交通运输工程—船舶及航道工程]