大工至善|大学至真分享 http://blog.sciencenet.cn/u/lcj2212916

博文

[转载]【电信学】【2004.11】基于神经网络的INS-GPS组合在陆地车辆导航中的应用

已有 151 次阅读 2021-2-19 16:27 |系统分类:科研笔记|文章来源:转载

图片


本文为加拿大卡尔加里大学(作者:Kai-Wei Chiang)的博士论文,共307页。

 

现代陆地车辆导航系统的大多数定位技术已经有25年的历史了。几乎所有的系统都扩充了这两种或更多的技术。组合导航系统的典型候选系统是全球定位系统(GPS)和惯性导航系统(INS)。Kalman滤波器已被广泛用作INS/GPS组合的最优估计工具,然而,这种多传感器组合方法的一些局限性已被报道,例如INS短期误差的影响模型依赖性、先验知识依赖性、传感器依赖性和线性化依赖性。为了减小惯导传感器短期误差的影响,通过频谱分析识别出运动动力学的带宽,采用了第一代离散小波变换(DWT)去噪算法来识别现有去噪算法的局限性。因此, 为了克服现有去噪算法的局限性,本研究提出了级联去噪算法。然后用多个INS/GPS组合陆地车辆系统对其进行了评价,结果表明,该方法在定位域和频谱域均优于现有的去噪算法。此外,还广泛研究了所提出算法对不同集成系统的影响。还提出了一种可替代的INS/GPS组合方法,即结合人工智能技术的概念智能导航器,以减少传统导航器使用Kalman滤波方法的局限性。提出的概念智能导航仪由多个不同的INS/GPS集成结构组成,这些结构是通过人工神经网络来获取导航知识的。此外,还实现了“大脑”、导航信息数据库和基于窗口的权值更新机制来存储和积累导航知识。利用多个INS/GPS组合导航系统对概念智能导航器进行了评估,结果表明,该导航器在位置域的性能优于传统导航器。最后,以一个低成本的INS/GPS组合导航系统为例,验证了将概念智能导航器作为开发下一代陆地车辆导航系统的替代方法所取得的优势。

 

Most of the positioning technologies for modern land vehicular navigation systems havebeen available for 25 years. Virtually all of the systems augment two or more of thesetechnologies. Typical candidates for an integrated navigation system are the Global Position System (GPS) and Inertial Navigation Systems (INS). The Kalman filter has been widely adopted as an optimal estimation tool for the INS/GPS integration, however, several limitations of such multi-sensor integration methodology have been reported; such as the impact of INS short term errors, model dependency, prior knowledge dependency, sensor dependency, and linearization dependency. To reduce the impact of short term INS sensor errors, the bandwidth of true motion dynamics were identified by spectrum analysis and the first generation denoising algorithm that used the Discrete Wavelet Transform (DWT) was applied to identify the limitations of the existing denoising algorithm.Consequently, this research proposed the cascade denoising algorithm to overcome the limitations of existing denoising algorithms. It was then evaluated using several INS/GPS integrated land vehicular systems and the results demonstrated superior performance to existing denoising algorithms in both the positioning and spectrum domains. In addition, the impact ofproposed algorithms on different integrated systems was investigated extensively. Furthermore, an alternative INS/GPS integration methodology, the conceptual intelligentnavigator incorporating artificial intelligence techniques, was proposed to reduce the remaining limitations of traditional navigators that use the Kalman filter approach. The proposed conceptual intelligent navigator consisted of several different INS/GPS integration architectures that were developed using artificial neural networks to acquire the navigation knowledge. In addition, the “brain”, a navigation information database, and a window based weight updating scheme were implemented to store and accumulate navigation knowledge. The conceptual intelligent navigator was evaluated using several INS/GPS integrated land vehicular systems and the results demonstrated superior performance to traditional navigator in the position domain. Finally, a low cost INS/GPSintegrated system was considered to verify the advantages gained by incorporating the conceptual intelligent navigator as an alternative method toward developing next generation land vehicular navigation systems.

 

1.       引言

2. 陆地车辆导航系统

3. INS/GPS集成基础

4. IMU信号的级联降噪

5. 人工神经网络的方法

6. 概念型智能导航器的研发

7. 结果与讨论

8. 结论与建议


更多精彩文章请关注公众号:205328s611i1aqxbbgxv19.jpg




http://blog.sciencenet.cn/blog-69686-1272917.html

上一篇:[转载]【电力电子】【2018.08】三相异步电动机的等效电路
下一篇:【课件】分享一份高频电子线路导论的课件,欢迎批评指正

0

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备07017567号-12 )

GMT+8, 2021-4-16 15:37

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部