摘要： 基于递归网络分离结构并利用时间相关的评价函数,针对二输入二输出盲信号分离问题,提出了一种非平稳信号的自适应盲分离算法。该算法计算量小,可根据输出信号能量大小有选择地更新分离系数,并可扩展到多输入多输出盲分离问题。仿真验证对声音等非平稳信号具有良好的分离效果。

关键词: 盲分离, 评价函数, 递归结构, 非平稳信号

Abstract: A double-input/double-output self-adaptive blind source separation (BSS) algorithm for non-stationary signals based on recursive structure and using time-related cost function is presented in this paper. This algorithm has low computing cost and could update selected separation parameters based on signal output energy which could be extended to multiple input/output blind source separation problem. Simulations have been made to validate the good separation performance of the algorithm.

中图分类号: 

• TP183