基于神经网络反演法的3mm波辐射特性研究  

Study on 3 Millimeter Wave Radiation by Using Neural Network Inversion Method

作　　者：龚冰[1] 娄国伟[1] 李兴国[1] 聂建英[1,2] 

机构地区：[1]南京理工大学电子工程与光电技术学院,南京210094 [2]福州大学数学与计算机科学学院,福州350002

出　　处：《科技导报》2009年第23期62-65,共4页Science & Technology Review

基　　金：国防预研基金项目(9140A01020107BQ0211)

摘　　要：被动探测系统的天线温度与目标视在温度的关系是一种病态积分方程。为了从第一类Fredholm积分方程中解出物体的视在温度,提出了一种利用天线温度反演视在温度的动态逼近方法。首先,利用3mm辐射计对不同条件下的目标及其背景进行测量,得出相应的天线温度。然后,引入辐射温度因子,并给出视在温度影响因素的权值,建立一种基于BP神经网络的视在温度动态逼近分析方法。最后,利用该方法分析了物理温度等因素对视在温度的影响,为毫米波辐射特性研究提供了一种有效的温度反演方法。由于天线温度可以通过测量直接得出,所以利用温度因子仿真曲线,可以比较快速地计算出变温目标的视在温度及其辐射特性信息。The relationship between the passive detecting system＇s antenna temperature and the apparent temperature can be expressed by a morbid integral equation. A dynamic approximation method for the inversion of the apparent temperature based on the antenna temperature is proposed, by solving a Fredholm integral equation of the first kind. The passive detector in this paper is 3 mm Dicke radiometer. Firstly, different objects and backgrounds are measured by the radiometer to obtain the antenna temperature. Secondly, withthe introduction of the factor of radiation temperature and the factor weight of the apparent temperature, the dynamic approximation method is formulated by using BP neural network. Finally, the effect factors on the apparent temperature are analyzed based on the above method. The relevant antenna temperature can be obtained by the measurement directly. The curve of temperature factors can be used to calculate the apparent temperature of objects easily to obtain the objects radiation characteristics. The above approximation method can be used in the study of millimeter wave radiation characteristics.

关 键 词：毫米波 反演 辐射特性 BP神经网络 辐射温度因子 

分 类 号：TP722[自动化与计算机技术—检测技术与自动化装置]