多层前向神经网络的RLS修正训练算法

A Modified RLS Algorithm for Multilayer Feedforward Neural Networks

提出了一种修正的基于递推最小二乘算法（ＲＬＳ） 的多层前向神经网络的快速学习算法，对两个系统进行了辨识，并与ＲＬＳ 训练算法和标准反向传播（ＢＰ）算法进行了比较。

A modified learning algorithm based on the recursive least squares(RLS)method for the multilayer feedforword neural networks is proposed. The algorithm has been applied to the identifications of two systems. The simulation results demonstrate the robustness of the system and the effectiveness of the proposed method.