论文部分内容阅读
提出了能够用于非线性系统建模的一种新型回归网络 ,该网络是Elman网络的改进 ,由输入层、隐层和输出层构成 .输入层由外部输入和内部状态层组成 ,隐层到状态层的连接是任意的 ,因此在逼近系统时 ,改进的Elman网络比Elman网络有更多记忆空间 .同时证明了改进的Elman网络能够逼近一定时间内的非线性系统的输出轨线 ,提出了利用动态反向传播算法训练神经网络的前向和反向权值 ,仿真结果验证了该方案的有效性
A new type of regression network which can be used for nonlinear system modeling is proposed, which is an improvement of Elman network and consists of input layer, hidden layer and output layer. The input layer is composed of external input and internal state layer, The connection of the state layer is arbitrary, so the improved Elman network has more memory space than the Elman network when approaching the system.At the same time, it is proved that the improved Elman network can approximate the output trajectory of the nonlinear system over a certain period of time, Using the dynamic backpropagation algorithm to train the forward and inverse weights of the neural network, the simulation results verify the effectiveness of the scheme