GA-BP神经网络的非线性函数拟合  

GA-BP Neural Network of the Nonlinear Function Approximating

作　　者：徐富强[1] 钱云[1] 刘相国[1] 

机构地区：[1]巢湖学院数学系

出　　处：《微计算机信息》2012年第7期148-149,145,共3页Control & Automation

基　　金：巢湖学院自然科学研究资助项目(XLY-201101)

摘　　要：由于BP神经网络具有良好的非线性映射能力,但在极值寻优时易陷入局部极值。本文利用具有全局搜索寻优的遗传算法优化BP神经网络(GA-BP),用于非线性函数拟合,弥补了BP网络寻优时的缺点。通过实例比较分析,GA-BP神经网络明显地提高了拟合精度,也使模型具有较强的应用性与推广性。Because of the BP neural network has good nonlinear mapping capabilities, but easily fall into the local extremum in optimization extreme. This paper use the genetic algorithm, which has global search optimization, to optimize the BP neural network （GA-BP）. Then used for nonlinear function fitting. These can remedy the BP network ’s shortcomings of optimizing. Through the example and comparative analysis, GA-BP neural network obviously improve the fitting precision, can also make the model with strong applicability and extension.

关 键 词：遗传算法 BP神经网络 函数拟合 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]