利用改进的BP算法实现神经网络辨识仿真  

Realization of Simulation for Neural Network Identification by Using Improved BP Algorithm

作　　者：攸立准 马飞[2] 徐海锋[3] 

机构地区：[1]衡水路桥工程有限公司,河北衡水053000 [2]中国电子科技集团公司第五十四研究所,河北石家庄050081 [3]石家庄铁道学院,河北石家庄050043

出　　处：《无线电工程》2009年第11期52-54,共3页Radio Engineering

摘　　要：系统辨识是控制系统设计的基础。基于多层前馈神经网络结构,采用一种改进的BP算法,利用二阶梯度变尺度模型,完成了神经网络非线性系统辨识。与传统的辨识方法比较,神经网络应用于非线性系统辨识具有泛化功能和很好的容错能力,是一种不依赖模型的自适应函数估计器。采用一种改进的BP算法有效地改善了系统收敛速度慢的问题,BP模型已成为神经网络的重要模型之一,从而为控制系统正确设计奠定理论基础。System identification is the basis of control system design. Compared with traditional identification method, neural network is a model-independent adaptive function estimator, which has good generalization and error tolerance performance when applied to nonlinear system identification. BP algorithm is a very efficient algorithm and BP model has become one of the important neural network models. By using an improved BP algorithm, the problem of slow convergence is improved efficiently. That has established a theoretical basis for control system design.

关 键 词：BP算法 神经网络 系统辨识 非线性系统 

分 类 号：TP387[自动化与计算机技术—计算机系统结构]