摘要
不同时间结构的平稳随机信号具有不同的模型结构,平稳随机信号可以由白噪声激励一自回归(AR)模型得到,在某种意义上AR模型与线性预测模型等价。因此,在盲源分离中线性预测模型可以作为度量信号分离的测度。为此从信号预测模型的角度出发分析推导了一种新的盲源分离算法,并进行了计算机仿真验证,实验结果表明该算法简单有效,便于工程实现。
Stationary random signals with different temporal structures often have different model structures.And this type of signals can be generated by the white noise through an auto-regressive（AR） model.Since in some sense,the AR model is equivalent to the linear prediction model,and the linear prediction model structure can be served as a measure to identify different signals.Based on the new measure,a novel blind source separation method is proposed and the corresponding simulation is carried out.Simulation results show that the new method is valid,efficient,and easy to implement in practice.
出处
《数据采集与处理》
CSCD
北大核心
2011年第2期162-166,共5页
Journal of Data Acquisition and Processing
基金
国家自然科学基金(61062002)资助项目
甘肃省自然科学基金(1014RJZA011)资助项目
关键词
盲源分离
信号模型
线性预测
瑞利熵
blind source separation
signal model
linear prediction
Raleigh entropy
