变结构神经网络及其应用

A Variable-structure Neural Network and Its Application

本文提出了变结构神经网络模型及其学习和工作算法。变结构神经网络可以有教师学习，也可以无教师学习，其学习过程没有迭代循环，权值一次确定，学习精度为０。数值仿真表明：对于非平稳环境，变结构神经网络皆有容错和弹性应变能力。

A variable-structure neural network was presented in this paper, including its learning and working algorithm. The neural network could learn with teacher or without teacher. The learning process was non-cycle and non-iterating. The weights were set up at a time. The neural network was trained without any error. Finally, the numerical simulation was given to show that the neural network could not only learn on the analysis of unanticipated features but also work under noisy conditions.