PINN-TurbNet for predicting atmospheric turbulence in LEO satellite-to-ground laser communication links

Atmospheric turbulence significantly affects the quality of laser communication links between low Earth orbit (LEO) satellites and ground stations. To enable accurate prediction of turbulence around ground stations, this study conducts a quantitative analysis of the communication link and presents the spatial distribution of atmospheric turbulence intensity. We propose a physics-informed neural network model, PINN-TurbNet, which integrates high-resolution meteorological parameters with physical constraints to enhance prediction accuracy. Furthermore, a multi-layer atmospheric turbulence analysis approach is introduced to improve computational efficiency. Experimental results demonstrate that the PINN-TurbNet model achieves a relative mean squared error (MSE) of 0.318 and a relative mean absolute error (MAE) of 0.209.

大气湍流（Atmospheric turbulence）会显著影响低地球轨道（low Earth orbit, LEO）卫星与地面站之间的激光通信链路质量。为实现地面站周边湍流的精准预测，本研究对通信链路展开定量分析，并给出了大气湍流强度的空间分布特征。本研究提出一种物理信息神经网络（physics-informed neural network, PINN）模型PINN-TurbNet，该模型融合高分辨率气象参数与物理约束以提升预测精度。此外，本研究还引入了多层大气湍流分析方法，以提升计算效率。实验结果表明，PINN-TurbNet模型的相对均方误差（mean squared error, MSE）为0.318，相对平均绝对误差（mean absolute error, MAE）为0.209。