一种采用径向基神经网络的盲源分离算法

摘要

针对传统盲源分离算法的计算复杂问题,提出一种基于径向基(RBF)神经网络盲源分离算法,用K均值聚类算法对中心值和宽度值进行确定,用最大熵为代价函数来确定权值,所用的代价函数保证了网络的输出尽可能独立,使信号能正确地分离.仿真中,用于对线性混合信号进行盲源分离,并与最大熵(ME)算法进行比较.结果表明,盲源分离算法能减少分离时间和提高分离效率,并且能大大降低计算量,比ME算法更好.%This paper presented a radial basis function(RBF) neural network algorithm for blind source separation. The value of central and width determined by K - means clustering algorithm. The weights were determined by the cost function of maximum entropy, which can ensures the output of the network as independent as possible and the signals can be separated correctly. The algorithm used Linear blind source separation of mixed - signals and compared with the maximum entropy( ME)algorithm in simulation experiment. The results show that the algorithm can reduce the separation time and improve the efficiency, and can greatly reduce the computation. The performances of the algorithm is better than ME algorithm.