基于反馈神经网络的动态化工过程建模

针对非线性动态化工过程建模存在的问题，提出了一种新的反馈神经网络结构，并将状态反馈、时间序列延尺以及集中节点的概念结合起来，用于提高反馈神经网络的性能，同时又使得网络结构不至于太复杂，在用此网络结构建模的时间，成功地将BP算法用一网络模型的训练。文中将这种反馈神经网络结构分别对一个单输入单输出（SISO）的非线笥动态系统和一个多输入单输出（SIMO）的连续全混釜（CSTR）模型进行建模，并将所得模型与基于表态BP神经神经所得的模型在模型输出精度和抗干扰性等方面进行了比较，证明了该反馈神经在动态过程建模中能够比静态BP模型更好地反映出动态过程的输入输出关系，并具有一定的抗干扰能力。

Modeling Nonlinear Dynamic Chemical Process Base on Artificial Neural Networks

A new recurrent neural network, based on dynamic characteristics of chemical process, is put forward in this article. The structures of state feedback, time delayed nodes and the integrated nodes are well combined in its structure so as to make the network memorize more past system states and keeping it from being too complex.The static BP algorithm is successfully used to train it. The new recurrent neural network is used to build models for a single\|input\|single\|output (SISO) system and a multi\|input\|single\|output (MISO) system and then the models are compared with other models based on BP neural networks.The comparison result shows multi layer feed forward neural networks. The result shows that models based on the new recurrent neural networks are more reliable and have high capability of antijamming.