摘要
在过程信号的去噪中,应用较新的盲信号神经网络分离(BSS)的方法,但盲信号分离神经网络存在容易陷入局部极小点、收敛速度慢的缺点。为此进一步采用蚁群算法(Ant Colony Algorithm,简称ACA)优化盲信号分离神经网络权值的初值,将蚁群算法与神经网络(HJNN)结合形成AC-HJNN算法,可迅速得到最佳盲信号分离神经网络的权值矩阵,实现对过程信号的去噪。仿真实验表明:用AC-HJNN算法,可兼有神经网络广泛映射能力和蚁群算法快速全局收敛的性能。
One of the main weak-points of the blind separation algorithm of HJNN is that the optimal procedure is easily stacked into the local minimal value,which causes slow convergence. Based on ant colony algorithm （ACA）, AC-HJNN algorithm is proposed to optimize the initial value of the weight of HJNN so as to obtain the optimum weight matrix quickly and realize the denoising of the process signal. The comparison between the two algorithms is given with experiment.
出处
《计算机应用与软件》
CSCD
北大核心
2007年第8期21-22,37,共3页
Computer Applications and Software
基金
上海市教委科技发展基金(050Z02)资助项目。
关键词
蚁群算法
神经网络
盲分离
过程信号
Ant colony algorithm Neural network Blind separation Process signal
