摘要
为了按照规范四阶累积量的绝对值有序分离出源信号,提出了一种新的基于粒子群优化的有序盲信号分离算法.本算法采用信号的规范四阶累积量作为代价函数,使用粒子群优化算法代替传统的梯度算法对代价函数进行优化,通过消源去相关方法从混合信号中消去已分离出的源信号成分,逐次按序提取出源信号,解决了梯度算法容易陷入局部极值而不能正确提取源信号的缺点.仿真结果表明,本算法能够有效实现对各类源信号的有序盲分离,且分离顺序能够确保按照源信号的规范四阶累积量绝对值的降序进行.
A new sequential blind signal separation algorithm based on particle swarm optimization was proposed for separating source signal according to the absolute value of normalized fourth-order cumulant. Normalized fourth-order cumulant was used in the algorithm as cost function and traditional gradient algorithm was replaced by particle swarm optimization algorithm for optimizing the cost function. The separated source signal component was wiped off using deflation method and source signal could be separated in sequence. In the meantime, the shortcoming of being easy for gradient algorithm to fall into local extremum was avoided. The simulation results show that the algorithm can achieve the efficient sequential blind separation for various types of source signal and ensure the separation order according to the descending absolute value of normalized fourth-order cumulant.
出处
《天津大学学报》
EI
CAS
CSCD
北大核心
2011年第2期174-179,共6页
Journal of Tianjin University(Science and Technology)
基金
国家自然科学基金资助项目(60802049)
中国博士后科学基金资助项目(20060390170)
天津市高校科技发展基金资助项目(20080710)
关键词
粒子群优化
有序盲分离
消源
规范四阶累积量
particle swarm optimization
sequential blind separation
deflation
normalized fourth-order cumulant
