基于回归神经网络的复杂工业对象的建模

Modeling for A Complicated Industrial Object Based on Recurrent Neural Network

摘要: 讨论一种动态神经网络——Elman回归神经网络的结构和算法.基于这一网络结构提出了非线性时变工业对象——直流电弧的神经网络建模方法,并与用其他方法为对象建立的模型进行了比较,结果证明回归网络模型能够很好地适配该工业对象,显示了动态神经网络在工业对象建模中的良好应用前景.

Abstract: The architecture and algorithm of a kind of dynamical neural network, Elman recurrent neural network (RNN)were dicussed. Based on this network, an approach of modeling for nonlinear time-varying industrial object, direct current arc, was proposed. Compared with other modeling method for the object, the model based on RNN is proved to have better performance.