National Time Service Center, Chinese Academy of Sciences, Xi'an 710600, China.
Key Laboratory of Precise Positioning and Timing Technology, Chinese Academy of Sciences, Xi'an 710600, China.
Sensors (Basel). 2023 May 29;23(11):5176. doi: 10.3390/s23115176.
The core of eLoran ground-based timing navigation systems is the accurate measurement of groundwave propagation delay. However, meteorological changes will disturb the conductive characteristic factors along the groundwave propagation path, especially for a complex terrestrial propagation environment, and may even lead to microsecond-level propagation delay fluctuation, seriously affecting the timing accuracy of the system. Aiming at this problem, this paper proposes a propagation delay prediction model based on a Back-Propagation neural network (BPNN) for a complex meteorological environment, which realizes the function of directly mapping propagation delay fluctuation through meteorological factors. First, the theoretical influence of meteorological factors on each component of propagation delay is analyzed based on calculation parameters. Then, through the correlation analysis of the measured data, the complex relationship between the seven main meteorological factors and the propagation delay, as well as their regional differences, are demonstrated. Finally, a BPNN prediction model considering regional changes of multiple meteorological factors is proposed, and the validity of the model is verified by long-term collected data. Experimental results show that the proposed model can effectively predict the propagation delay fluctuation in the next few days, and its overall performance is significantly improved compared with that of the existing linear model and simple neural network model.
基于 eLoran 的地面定时导航系统的核心是对地面波传播延迟的精确测量。然而，气象变化会干扰地面波传播路径沿线的导电特征因素，特别是对于复杂的地面传播环境，甚至可能导致微秒级的传播延迟波动，严重影响系统的定时精度。针对这个问题，本文提出了一种基于反向传播神经网络（BPNN）的复杂气象环境下传播延迟预测模型，该模型通过气象因素实现了直接映射传播延迟波动的功能。首先，基于计算参数分析了气象因素对传播延迟各分量的理论影响。然后，通过对实测数据的相关分析，揭示了七种主要气象因素与传播延迟之间的复杂关系及其区域差异。最后，提出了一种考虑多个气象因素区域变化的 BPNN 预测模型，并通过长期采集的数据验证了模型的有效性。实验结果表明，所提出的模型可以有效地预测未来几天的传播延迟波动，与现有的线性模型和简单神经网络模型相比，其整体性能有了显著提高。
