基于Cornell的自适应电离层闪烁强度的模型研究

针对Cornell模型在同一闪烁条件下,数据更新周期越长,估计和模型输入的幅度闪烁指数S4的偏差越大,并随着电离层闪烁的增强而增强的问题,提出基于Cornell模型的自适应S4的电离层闪烁模型即AS4-Cornell模型。模型以S4的偏差作为BP(back propagation)神经网络比例积分微分(proportional integral derivative, PID)算法的反馈,自动调整输入的复高斯白噪声的区段权值,使最终产生的闪烁信号满足模型输入的闪烁指数S4的指标。结果表明：仿真时,AS4-Cornell模型的幅度和相位闪烁序列概率分布均符合电离层闪烁理论,估算得到的电离层幅度指数S4与模型输入S4指数的最大偏差为0.001;全球定位系统(global positioning system, GPS)电离层闪烁模拟器测试时,AS4-Cornell模型估算得到的电离层幅度指数S4与模型输入S4指数的最大偏差为0.09;相比Cornell模型,AS4-Cornell模型产生的电离层闪烁信号更能够很好地反映模型输入的电离层闪烁指数S4的强度。

A model study of adaptive ionospheric scintillation intensity based on Cornell

For Cornell model in the same scintillation conditions, the longer the data update period, the greater the deviation of the estimated and model input amplitude scintillation index S4, and with the enhancement of ionospheric scintillation, the ionospheric scintillation model based on Cornell model with adaptive S4 is proposed, namely AS4-Cornell model. The model uses the deviation of S4 as the feedback of the BP neural network PID algorithm to automatically adjust the zone weights of the input complex Gaussian white noise so that the final generated scintillation signal meets the index of the scintillation index S4 input to the model. The results show that the amplitude and phase scintillation sequence probability distributions of the AS4-Cornell model are consistent with the ionospheric scintillation theory during the simulation, and the maximum deviation of the estimated ionospheric amplitude index S4 from the model input S4 index is 0.001; the maximum deviation of the estimated ionospheric amplitude index S4 of the AS4-Cornell model from the model input S4 index during the GPS ionospheric scintillation simulator test is 0.001. The maximum deviation of the ionospheric amplitude index S4 estimated by the AS4-Cornell model from the model input S4 index is 0.09; compared with the Cornell model, the ionospheric scintillation signal generated by the AS4-Cornell model can better reflect the intensity of the ionospheric scintillation index S4 inputted by the model.