摘要
在水声信号盲分离技术仿真研究中 ,需要评价算法的收敛速度 ,稳态误差等。文中分析了水声基阵接收信号的实数延时模型和复数混合模型 ,提出了水声信号盲分离算法性能评价准则。该准则分为基于混合基阵和信号估计两种。在三 /四元均匀线列阵上 ,盲分离实测非高斯舰船辐射噪声 ,分别用 CRLS、CACY、EASI、CMA、MKMA算法验证了提出的算法评价准则 ,同时 ,还给出了这些算法分离水声信号的性能曲线。
In computer simulation for blind signal separation, we need to evaluate the separating algorithms, for example, convergence rate, steady state error, and so on. According the real domain algorithms, we present the evaluating criterion to the complex algorithms for blind underwater acoustic signals separation. The criterion is divided into two parts based on mixing matrix and source signals. RLS, ACY, EASI, CMA, MKMA algorithms are used to separate the non Gaussian ship noises recorded in computer simulation with 3/4 elements uniformly linear array. The simulation curves are given. Comparing several criterions based mixing matrix estimation, we present the optimal criterion.
出处
《探测与控制学报》
CSCD
北大核心
2002年第2期40-44,共5页
Journal of Detection & Control
基金
国家自然基金资助的项目 (60 0 72 0 5 2 )
关键词
水声信号
盲分离
水声基阵
算法评价
声纳
blind underwater acoustic signal separation
underwater acoustic array
algorithms evaluation
