摘要
预警雷达探测过程中气动目标微动回波能量弱导致识别性能不稳定。针对该问题,提出一种基于稀疏约束非负矩阵分解(sparse constrained non-negative matrix factorization,SCNMF)和集成极限学习机(integrated extreme learning machine,IELM)的多频点调制谱融合增强识别方法。通过分析微动部件回波特性,对多频点频域幅度谱进行SCNMF处理实现像素级融合得到特征增强后的稀疏调制谱,并将其作为样本输入IELM,实现气动目标类型识别。仿真和实测数据表明,本文方法能够有效融合多频点微动特征,具有抗噪能力强、所需训练样本少和识别性能稳健等优势。
To solve the problem of unstable recognition performance caused by weak energy of aircraft fretting echo in early warning radar detection process,a multi-frequency modulation spectrum fusion and enhanced recognition method combining sparse constrained non-negative matrix factorization(SCNMF)and integrated extreme learning machine(IELM)is proposed.By analyzing the echo frequency domain characteristics of the micro-motion parts,SCNMF is performed on the modulation spectrum of multi-frequency to achieve pixel-level fusion and obtain the enhanced sparse modulation spectrum,which is input into IELM as a sample to achieve pneumatic target classification.Simulation and measured data verify that the proposed method can effectively integrate multi-frequency micro-motion features,and has the advantages of strong anti-noise ability,fewer training samples and robust recognition performance.
作者
赵庆媛
赵志强
叶春茂
鲁耀兵
ZHAO Qingyuan;ZHAO Zhiqiang;YE Chunmao;LU Yaobing(Beijing Institute of Radio Measurement,Beijing 100854,China)
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2023年第7期2043-2050,共8页
Systems Engineering and Electronics
基金
国家自然科学基金(60831001)
国防基金(9140A31010109HK0101)资助课题。
关键词
调制谱
气动目标
稀疏约束非负矩阵分解
集成极限学习机
modulation spectrum
pneumatic target
sparse constrained non-negative matrix factorization
integrated extreme learning machine(IELM)
