...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Journal of guidance, control, and dynamics >Newton–Cotes Discretization for Improved Dead-Reckoning in Bayesian Estimators with Limited Sampling Rate
【24h】

Newton–Cotes Discretization for Improved Dead-Reckoning in Bayesian Estimators with Limited Sampling Rate

机译：牛顿-科茨离散化技术可改善采样率有限的贝叶斯估计中的死区

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The current work has presented a novel Newton–Cotes-baseddiscretization approach capable of reducing dead-reckoning errorfor Bayesian estimators in applications where the system is not fullyobservable. This is achieved by reducing numerical integration error,an often-overlooked contributor to dead-reckoning error. The effectivenessof the approach was demonstrated in practice through applicationto a satellite attitude determination problem, where it wasfound that the Newton–Cotes-based approach outperformed thetypical zero-order hold approach approximately 94% of the time.The new approach improves over past approaches to addressingnumerical integration error, such as Runge–Kutta integration, in thatit does not rely on the zero-order hold assumption for internalmeasurements. Instead, it leverages the past behavior of the motionmodel to predict its future behavior, resulting in a more accurateprediction of the state. For this reason, the Newton–Cotes approach ishighly effective in problems involving highly dynamic motions.Furthermore, the approach is able to achieve better performancewithout the need for any changes to hardware architecture or additionalsensor data. Possible extensions of the current work include theuse of quadrature rules that are robust to numerical stability issuesand uneven sampling intervals.
机译：当前的工作提出了一种新颖的基于牛顿-科特斯的离散化方法，该方法能够为无法完全观测到系统的应用中的贝叶斯估计器减少死区重收误差。这是通过减少数值积分误差来实现的，数值积分误差经常被认为是死区推算误差的原因。在实践中通过应用卫星姿态确定问题证明了该方法的有效性，发现基于牛顿-科特斯的方法在大约94％的时间内胜过典型的零阶保持方法。新方法比过去解决数字方法的方法有所改进积分误差，例如Runge-Kutta积分，因为它不依赖内部测量的零阶保持假设。相反，它利用运动模型的过去行为来预测其未来行为，从而对状态进行更准确的预测。因此，Newton-Cotes方法在涉及高动态运动的问题上非常有效。此外，该方法无需改变硬件体系结构或其他传感器数据即可实现更好的性能。当前工作的可能扩展包括使用正交规则，该规则对于数值稳定性问题和不均匀的采样间隔具有鲁棒性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号