基于EFA的声呐混响抑制两级降维算法研究  

A research on two-stage dimensionality reduction algorithm for sonar reverberation suppression bas-ed on EFA

作　　者：梁浩前 王晓庆[1] 王大宇[1] LIANG Hao-qian;WANG Xiao-qing;WANG Da-yu(The 54th Research Institute of CETC,Shijiazhuang 050081,China)

机构地区：[1]中国电子科技集团公司第五十四研究所,石家庄050081

出　　处：《信息技术》2023年第5期55-61,共7页Information Technology

基　　金：国家自然科学基金资助项目(U19B2028)。

摘　　要：针对传统的空时自适应处理降维手段如扩展因子法、局域联合处理法,在大阵列条件下存在计算量大、实时性差和空域混响抑制效果差等问题,提出一种两级降维算法。该算法在扩展因子法第一级降维基础上,通过一对二次代价函数循环迭代进行第二级降维。仿真结果表明,该算法相比扩展因子法所需计算量与训练样本大幅减少,相比局域联合处理法,降低了波束主瓣展宽,空域混响抑制效果更好。Regarding the problems such as large amount of calculation,poor real-time performance and poor spatial reverberation suppression ability under the condition of large arraysof traditional space-time adaptive processing dimensionality reduction algorithms including Extended Factored Approach(EFA)and Joint Domain Localized(JDL),a two-stage dimensionality reduction algorithm is proposed.Based on the first-stage dimensionality reduction of EFA,this algorithm performs the second-stage dimensionality reduction through loop iterations of a pair of quadratic cost function.The simulation results show that compared with EFA,the required calculation and training samples are greatly reduced by this algorithm,and compared with JDL,the main lobe broadening of the beam is reduced and the spatial reverberation suppression effect is better.

关 键 词：混响抑制 空时自适应处理 降维 主动声呐 混响协方差矩阵 

分 类 号：TN911.7[电子电信—通信与信息系统]