基于人工神经网络的毫米波雨衰减预报

Rain Attenuation Prediction of Millimeter Wave Based on Artificial Neural Network

在考虑了电波雨衰减与工作频率、仰角、降雨率、极化角等多个参数的综合非线性映射的基础上 ,建立了基于人工神经网络的毫米波雨衰减预报模型 ,并将结果与 CCIR预报模型进行了分析比较 .结果表明 ,利用神经网络的非线性特性进行高频电波雨衰减的预报 ,可降低平均误差 0 .59d B,并减小均方差 0 .69d

Based on the consideration of nonlinear mapping between rain attenuation and compositions of various parameters such as frequency, elevation angle, rain-fall rate, polarization angle and so on, this paper proposed a new rain attenuation predication model of millimeter wave. The predicted results of the model were compared with that of the CCIR model. The results show that applying the artificial neural network to predict rain attenuation of high frequency wave decreases the mean prediction error by 0.59 dB and the root of mean square error by 0.69 dB.