基于多分类支持向量机的相控阵雷达T/R组件寿命分布仿真识别  

Identification of T/R Module's Life Distribution of Phased Array Radar with Multi-Class Support Vector Machine

作　　者：蒋伟[1] 王挺 盛文[1] 鲁力[1] JIANG Wei;WANG Ting;SHENG Wen;LU Li(Air-Defense Early Warning Equipment Department,Air Force Early Warning Academy,Wuhan 430019,China)

机构地区：[1]空军预警学院防空预警装备系,武汉430019

出　　处：《兵器装备工程学报》2018年第11期89-93,共5页Journal of Ordnance Equipment Engineering

基　　金：军内科研重点项目(KJ2012225)

摘　　要：为了能够对T/R组件的寿命分布类型进行快速准确地识别,在充分分析T/R组件寿命分布数据特点的基础上,建立了核主元降维-多分类支持向量机识别模型。首先对T/R组件的数据特征以及可能的分布模式进行选取,其次利用Matlab仿真软件产生100组不同分布的随机数,构建了模型的初始训练样本,最后利用核主元降维分析法的非线性主元特征提取能力以及多分类支持向量机模型的高精度识别能力,对T/R组件的寿命分布类型进行识别。实例计算表明,该识别模型具有较高的识别精度,与传统极大似然估计和k-s检验法的T/R组件寿命分布结果一致,从而证实了模型的正确性,为后续T/R组件的维修提供理论依据。In order to be able to identify the type of T/R modules’life distribution quickly and accurately,by fully analyzing the T/R modules’characteristics of life distribution data,the kernel principal component dimensionality reduction and multi-class support vector machine model was established.First of all,T/R modules’data features and possible distribution patterns were selected,then100groups of random number with the different distribution were generated by the Matlab simulation software,and an initial training sample of the model was constructed,and with making full use of the nonlinear feature extraction of kernel principal component dimensionality reduction and high-precision identification capabilities of multi-class support vector machine,the life distribution of T/R modules was identified.The calculation results show that the precision and stability of the model are proved,and the life distribution results are consistent with the maximum likelihood estimation and k-s test method,thus verifying the correctness of the model and providing theoretical basis for the maintenance of T/R modules.

关 键 词：T/R组件 寿命分布 核主元降维 支持向量机 k-s检验 

分 类 号：TN95[电子电信—信号与信息处理]