摘要
提出了一种新的模糊神经网络的快速参数学习算法 ,采用一些特殊的处理 ,可以用递推最小二乘法 (RLS)来调整所有的参数 .以前的学习算法在调整模糊隶属度函数的中心和宽度的时候 ,用的是梯度下降法 ,具有容易陷入局部最小值点、收敛速度慢等缺点 ,而本算法则可以克服这些缺点 ,最后通过仿真验证了算法的有效性 .
A novel parameter learning algorithm for fuzzy neural networks (FNN) is proposed. The conventional methods usually use the gradient descent based backpropogation algorithm to adjust the center and width of the membership functions. To avoid the inborn problem of BP algorithm, such as local minima and slow convergence, a modified RLS method is employed here to adjust the parameters of FNN, which is faster than the conventional BP algorithm. The validity of this method has been demonstrated by simulation results.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2002年第4期583-587,共5页
Control Theory & Applications
