Scholarlink

<p>Today, the importance of navigation accuracy is increasing due to its applications in intelligent transportation systems. The integrated navigation system, consisting of both INS and GPS systems, offers reliable, accurate, and continuous navigation capabilities compared to an independent INS or GPS. In this research, a new integrated navigation algorithm, Sage-Husa AKF + OI-SVR, is introduced to enhance accuracy and improve integrated navigation performance, especially during GPS signal outages. This algorithm is based on online incremental learning support vector regression alongside a modified Sage-Husa adaptive Kalman filter. In this paper, a concept of self-adaptation is introduced to the classic Kalman filter, and a modified Sage-Husa adaptive Kalman filter algorithm is developed based on a recursive noise estimator based on maximum a posteriori estimation. This aims to overcome the shortcomings of classic Kalman filter methods and address the state estimation problem in practical environments with complex noise, uncertain statistical characteristics, and model uncertainty. Many studies have been conducted on the possibility of GPS signal outages during the navigation process, but most of these methods are offline and cannot to manage data on a large scale, as they are costly in terms of memory consumption and computational volume. In addition, offline methods do not incrementally update the model with new information when the system gradually receives new data. Therefore, in this paper, support vector regression with online incremental learning is used to build a high-accuracy prediction model when GPS is functioning well and to provide the necessary observations for updating the adaptive Kalman filter during GPS signal outages. Real-world data from the sea trial was used to evaluate the proposed algorithm. The simulation results for field test data during GPS signal outages showed that the proposed integrated algorithm has a higher estimation accuracy compared to previous studies, managing to improve the standard deviation criteria for position and speed by 77%-46% and 96%-91%, respectively. </p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An online incremental learning support vector regression for INS/GPS integrated navigation system during long-time GPS outage

  • 期刊论文

  • 发表日期：

    2025年7月8日

查看全文
  • Seyed Mostafa Hosseini,
  • Mohamadreza Jalili

摘要

Today, the importance of navigation accuracy is increasing due to its applications in intelligent transportation systems. The integrated navigation system, consisting of both INS and GPS systems, offers reliable, accurate, and continuous navigation capabilities compared to an independent INS or GPS. In this research, a new integrated navigation algorithm, Sage-Husa AKF + OI-SVR, is introduced to enhance accuracy and improve integrated navigation performance, especially during GPS signal outages. This algorithm is based on online incremental learning support vector regression alongside a modified Sage-Husa adaptive Kalman filter. In this paper, a concept of self-adaptation is introduced to the classic Kalman filter, and a modified Sage-Husa adaptive Kalman filter algorithm is developed based on a recursive noise estimator based on maximum a posteriori estimation. This aims to overcome the shortcomings of classic Kalman filter methods and address the state estimation problem in practical environments with complex noise, uncertain statistical characteristics, and model uncertainty. Many studies have been conducted on the possibility of GPS signal outages during the navigation process, but most of these methods are offline and cannot to manage data on a large scale, as they are costly in terms of memory consumption and computational volume. In addition, offline methods do not incrementally update the model with new information when the system gradually receives new data. Therefore, in this paper, support vector regression with online incremental learning is used to build a high-accuracy prediction model when GPS is functioning well and to provide the necessary observations for updating the adaptive Kalman filter during GPS signal outages. Real-world data from the sea trial was used to evaluate the proposed algorithm. The simulation results for field test data during GPS signal outages showed that the proposed integrated algorithm has a higher estimation accuracy compared to previous studies, managing to improve the standard deviation criteria for position and speed by 77%-46% and 96%-91%, respectively.