摘要
针对信号源入射角时变的情况,分析了基于幂迭代的子空间跟踪算法,提出一种次分量分析方法的子空间跟踪算法。该算法首先利用基于反Hebbian学习的神经网络,抽取多个次分量,操作相对简单、算法稳定、收敛快,且有自组织特性;然后提出一种实时并行处理方法,在抽样结束时数据处理完成;最后采用牛顿法实现运动目标的DOA跟踪。仿真实验证明其收敛快、跟踪性能好。
As the direction of signal source varies with time,this paper proposed a tracking method based on minor component analysis approach.To extract the minor component,the algorithm was based on anti-Hebbian learning neural network and contain only relatively simple operations,it was stable,converges and had self-organizing properties.Then proposed a method of real-time parallel processing,at end time of sampling data,processing also could be finished.Finally,using Newton algorithm to track moving target.Computer simulation results demonstrate the effectiveness of the proposed algorithm.
出处
《计算机应用研究》
CSCD
北大核心
2010年第7期2492-2493,2519,共3页
Application Research of Computers
基金
国家"863"计划资助项目(2009AA7034530)
