多层前向神经网络的快速学习算法及其应用

High speed learning algorithm for a multilayer feedforward neural network and its application

针对目前多层前向神经网络学习算法存在的不足 ,提出一种多层前向神经网络的快速学习算法 ,它不仅符合生物神经网络的基本特征 ,而且算法简单 ,学习收敛速度快 ,具有线性、非线性逼近精度高等特性。以二杆机械手逆运动学建模作为应用实例 ,仿真结果表明该方法是有效的 ,其算法与收敛速度更优于 BP网络。

Akindofhighspeedlearning algorithm for a multilayer feedforward neural network is presented. It not only fits the basic characteristics of the biological neural network but also is a simple algorithmwithhighspeedlearningprocessand excellent characteristics in the linear or nonlinear accurate approximation. An application example of inverse kinematic model for a two bar robotic hand is given. The simulation results show the efficiency of the proposed algorithm.