### 典型隐身飞机的RCS起伏统计特性

1. 北京航空航天大学 航空科学与工程学院, 北京 100191
• 收稿日期:2013-12-30 修回日期:2014-04-14 出版日期:2014-12-25 发布日期:2014-05-07
• 通讯作者: 姬金祖 男, 博士, 讲师.主要研究方向: 飞行器隐身设计,计算电磁学,隐身材料. Tel: 010-82317503 E-mail: jijinzu@buaa.edu.cn E-mail:jijinzu@buaa.edu.cn
• 作者简介:陈世春 男, 博士研究生.主要研究方向: 飞行器总体设计,飞行器隐身设计. Tel: 010-82317503 E-mail: 36050225csc@sina.com;黄沛霖 男, 博士, 副教授.主要研究方向: 飞行器总体设计,飞行器隐身设计,计算电磁学. Tel: 010-82317503 E-mail: peilin_h@126.com

### Radar Cross Section Fluctuation Characteristics of Typical Stealth Aircraft

CHEN Shichun, HUANG Peilin, JI Jinzu

1. School of Aeronautic Science and Engineering, Beihang University, Beijing 100191, China
• Received:2013-12-30 Revised:2014-04-14 Online:2014-12-25 Published:2014-05-07

Abstract:

Radar cross section (RCS) fluctuation characteristics are used to predict radar's detection performance and evaluate aircraft's scattering characteristic. Azimuth RCS statistics under different incident conditions for six typical stealth aircraft at eight different bands each with two kinds of polarizations are obtained utilizing high-frequency electromagnetic computation method. Three kinds of relatively new fluctuation models are used to fit those statistics and some universal conclusions are made by analyzing all the fitting patterns. The Chi-square model gives a better peak value estimation for the probability density distribution curve but doesn't fit very well after the curve peak; the log-normal model often gives a higher estimation for the peak value but fits well after the curve peak; the fitting error of Chi-square model decreases when the double-degrees of freedom increases (to about 1.1-1.5), which results from the increasing ratio (about 0.1-1.0) of the mean square to the variance of the statistics, and according to the Kolmogorov-Smirnov testing method, the error is about 0.15 to 0.25; the fitting error of log-normal model also decreases (to about 1.5-5.0) when the ratio of mean to median value of the statistics decreases. When the statistics unit is dB·m2, the log-normal model and Legendre polynomials model always fit the data well more widely and the error is generally lower than 0.10.