复杂边界电波传播损耗的神经网络估计方法  

Modeling electromagnetic wave propagation loss using neural network

作　　者：樊琦[1] 高火涛[1] 王高峰[2] 

机构地区：[1]武汉大学电子信息学院,湖北武汉430079 [2]武汉大学微电子与信息技术研究院,湖北武汉430072

出　　处：《电波科学学报》2008年第1期184-188,共5页Chinese Journal of Radio Science

基　　金：国家高科技发展计划资助项目(2001AA631050);国家自然科学基金项目(90307017)

摘　　要：针对抛物方程用于模拟复杂地面边界电波传播计算时间随着频率升高急剧增加的缺点,提出利用抛物方程给神经网络提供训练数据,用电磁波频率和水平方向的距离作为神经网络的输入,传播损耗作为神经网络输出的电波传播损耗的分析估计方法。与传统抛物方程的计算结果对比表明,所提出的复杂边界电波传播损耗的估计方法,在保证了较高计算精度的同时,计算速度得到很大的提高。When the electromagnetic wave propagation loss is simulated by using parabolic equation, the computation time increases rapidly with the antenna frequency. A new neural network approach is presented to solve this problem. The training data of the neural network is obtained from the parabolic equation, the antenna frequency and the horizontal distance is used as the input of the neural network, and the output of the the neural network is considered as the propagation loss. Comparied with the traditional parabolic equation, the simulation results for several test frequency show that the trained neural network dramatically improved the computation speed without lossing much precision.

关 键 词：神经网络 抛物方程 电波传播 传播损耗 

分 类 号：TN011[电子电信—物理电子学]