权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Solution of Inverse Problems in Electromagnetic and Optical Propagation Using Artificial Neural Networks

使用人工神经网络解决电磁和光传播中的反问题

基本信息

批准号：
9014243
负责人：
Leung Tsang
金额：
$ 15.1万
依托单位：
University of Washington
依托单位国家：
美国
项目类别：
Continuing Grant
财政年份：
1991
资助国家：
美国
起止时间：
1991-03-01 至 1993-08-31
项目状态：
已结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=9014243&HistoricalAwards=false
关键词：
Solution Inverse Problems Electromagnetic Optical

项目摘要

In electromagnetic and optical propagation analysis, the forward problem denotes the calculation of fields and waves given the parameters of the media. The inverse problem is to calculate the target or media parameters from given or measured fields and waves. In the past, study of the forward problem has been done extensively. However there is a relative lack of systematic investigation of the electromagnetic and optical inverse problems because of the highly nonlinear relations between fields and the media parameters. With the use of recent Artificial Neural Networks (ANN), we have been successful in developing efficient and speedy algorithms to solve nonlinear inverse problems in fields and waves. Specifically we propose to apply and develop ANN algorithms and query learning techniques for the study of the following: 1. Inverse problems in microwave and optical scattering in remote sensing of agriculture and ice and snow will be investigated. Multiple scattering effects of discrete random media and rough surfaces will be included. The ANN algorithms and query learning techniques will be applied to both synthetic data as well as satellite and aircraft remote sensing measurements. 2. Fast acquisition of design parameters in the parametric frequency selective surface designs based on neural network inversion will be investigated. Constrained inversion for iterative acquisition of 2D/3D surfaces using bitmap representations will be included. These techniques will also be applied to identify new surface elements that will provide unconventional frequency responses.

在电磁和光传播分析中，正问题是指在给定介质参数的情况下对场和波的计算。反问题是根据给定或测量的场和波计算目标或介质参数。在过去，人们对正问题进行了广泛的研究。然而，由于场与介质参数之间的高度非线性关系，对电磁场和光学逆问题的系统研究相对较少。利用最新的人工神经网络(ANN)，我们已经成功地开发出高效、快速的算法来解场和波中的非线性反问题。具体地说，我们建议应用和发展人工神经网络算法和查询学习技术来进行以下研究：1.将研究农业和冰雪遥感中的微波和光学散射的反问题。离散随机介质和粗糙表面的多重散射效应将被包括在内。人工神经网络算法和查询学习技术将应用于合成数据以及卫星和飞机遥感测量。2.研究了基于神经网络逆的参数频率选择曲面设计中设计参数的快速获取。将包括使用位图表示法迭代获取2D/3D曲面的约束反演。这些技术还将用于识别将提供非常规频率响应的新表面元素。