基于模糊自适应变权重算法的函数链神经网络预测方法

来源期刊:中南大学学报(自然科学版)2011年第9期

论文作者:罗周全 左红艳 王爽英 王益伟

文章页码:2812 - 2818

关键词:函数链神经网络;模糊自适应变权重算法;预测;模糊;神经网络

Key words:functional link neural network; fuzzy adaptive variable weight algorithm; forecast; fuzzy; neural network

摘    要:为提高复杂工业系统非线性时间序列预测精度,将工业系统非线性时间序列不同的单个预测模型预测值作为函数链神经网络的原始输入值,并将原始输入值按正交的三角函数扩展得到的数值作为函数链神经网络扩展输入值,在分析函数链神经网络拟合充要条件的基础上,结合模糊自适应变权重算法计算函数链神经网络权重,建立基于模糊自适应变权重算法的函数链神经网络预测模型。研究结果表明:基于模糊自适应变权重算法的函数链神经网络预测方法的预测精度较高,并且平均误差和预测平方根误差均较小,具有较强的泛化能力;该模糊自适应变权重函数链神经网络预测模型可用于复杂非线性工业系统决策。

Abstract:

In order to enhance forecasting precision problem about nonlinear time series in complex industry system, a functional link neural network forecasting model was established based on fuzzy adaptive variable weight algorithm by using of making some forecasting values from different single forecasting models of nonlinear time series in complex industry system as original input values of functional link neural network, making the original input values as patulous input values of functional link neural network after the original input values being extended according to orthogonal trigonometric function, analyzing necessary and sufficient conditions of functional link neural network fitting and calculating the weight of functional link neural network based on fuzzy adaptive variable weight algorithm. The simulation analysis results and forecasting results of the severe harm rate in some mine reveal that the functional link neural network forecasting method based on fuzzy adaptive variable weight algorithm has higher accuracy than that of every single combined forecasting model or other forecasting model. The functional link neural network forecasting is of good extensive capability and is very useful for requirement decision in complex industry system.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号