摘要
将误差反馈原理引入神经网络的训练和使用,从而提出一种带输入补偿的多层前馈网络模型.将该模型用于非线性系统建模时,能有效消除模型的固有误差,补偿模型工作时出现的动态误差,提高建模精度.给出了新网络模型的结构、学习算法及工作方式,最后通过仿真试验证明了新模型在系统建模中的有效性。
This paper puts forward a new neural networks model with compensating inputs by applying the principle of error feedback in neural networks.Using this neural networks model for nonlinear systems modelling,the dynamic error can be effectively reduced.The structure,learning algorithm and working way of this new model are given.The simulation shows the neural networks model with compensating inputs is effective for nonlinear systems modelling.
出处
《信息与控制》
CSCD
北大核心
1996年第2期65-69,共5页
Information and Control
关键词
输入补偿
神经网络
非线性系统
控制理论
multi-layer networks
system modelling
compensating inputs
learning algorithm
training of networks
