论文部分内容阅读
针对复值信号的源数估计和有序分离等关键技术,提出一种基于人工蜂群优化的源数未知的复值盲源分离方法,该方法首先利用交叉互验技术来估算复数源信号的个数,然后通过人工蜂群算法优化峰度的绝对值来获得最佳分离向量,并实现了逐次恢复源信号的目的.仿真实验结果表明,该方法不仅能依峰度绝对值的降序实现服从任何分布源信号的盲分离,同时比其他方法具有更优越的估计性能.另外,提出一种基于峰度的欠定复盲源分离算法,该算法根据信号的统计特性构造了用于欠定混合情况下盲抽取向量的代价函数,然后通过人工蜂群算法优化其函数来获得最佳分离向量,通过多次分离来实现欠定复盲源分离的目的.通过对混合分布类型的复值源信号欠定盲分离仿真实验验证了该算法的有效性.
Aiming at the key techniques of source number estimation and ordered separation of complex valued signals, a new method of complex-valued blind source separation based on artificial bee colony optimization is proposed. This method first uses cross-mutual technique to estimate complex source signals And then optimize the kurtosis absolute value by artificial bee colony algorithm to get the best separation vector and achieve the purpose of recovering the source signal one by one.The simulation results show that this method not only can obey the descending order of kurtosis absolute value The blind separation of any distributed source signals and the superior performance of other methods are also proposed.In addition, an underdetermined complex blind source separation algorithm based on kurtosis is proposed, which is based on the statistical properties of signals, Blind vector extraction, and then optimize the function by artificial bee colony algorithm to obtain the best separation vector, and achieve the purpose of separating underdetermined complex blind source through multiple separation.By combining the complex distribution source signal Underdetermined blind separation simulation experiments verify the effectiveness of the algorithm.