摘要
针对短时交通量的高度非线性,提出一种基于遗传-小波神经网络的预测方法.该方法以前馈多层感知器的神经网络拓扑结构为基础,选择Morlet母小波基函数作为隐含层激活函数,以最简化结构概念进行网络泛化,并将误差反向传播,经遗传算法对网络连接权值修正.实例证明,该方法预测精度高,预测速度较快,能够满足实际工程的要求.
In the report, aimed at advanced nonlinearity of short-term traffic volume, a kind of forecasting meth- od named genetic-wavelet neural network was proposed, which is based on topological structure of multilayer feed-forward perceptions （MLPs）, and in which Morlet mother wavelet function is selected to be active function of hide-layer. The network was generalized by minimization feature structure （MFS） concept, and output error was back propagated to genetic algorithm module to optimize connection weights of network. The forecasting ex- ample indicated that the accuracy ~,d ,~ .c .t. 1_ ~
出处
《海南大学学报（自然科学版）》
CAS
2014年第1期55-59,共5页
Natural Science Journal of Hainan University
基金
国家自然科学基金(61304210)
福建工程学院科研发展基金(GY-211057)
福建省教育厅A类基金(JA11192)
关键词
短时交通量
小波
神经网络
遗传算法
short-term traffic volume
wavelet
neural network
genetic algorithm
