摘要
交通流量预测是智能交通系统的核心内容,系统中多个功能的实现都是以其为基础。针对城市路网中交通流量的时域性以及准周期特性,提出了一种基于改进小波神经网路算法的交通流量预测方法。利用具有时域分辨能力的小波神经网络对流量信号进行分类,以实现对交通流量的预测;采用加动量项的方法对网络权值及参数进行修正,避免了神经网络训练时收敛缓慢以及陷入局部极小。通过仿真实验验证,提出方法可实现对交通流量的准确预测,并且可以有效地提高网络学习率。
Traffic flow prediction is the core content of ITS,and the realization of many functions in the system is based on it.Aiming at the time domain and quasi-periodic characteristics of traffic flow in urban road network,a traffic flow prediction method based on improved wavelet neural network algorithm is proposed.Wavelet neural network with time-domain resolution is used to classify traffic signals in order to achieve traffic flow prediction.The weights and parameters of the network are modified by adding momentum term to avoid slow evolution and falling into local minimum in the training of the neural network.The simulation results show that the proposed method can accurately predict the traffic flow and effectively improve the network learning rate.
作者
肖维
高谦
XIAO Wei;GAO Qian(Shenzhen University,Shenzhen 518000;BOSCH Rexroth Co.,Ltd.,Changzhou 213000)
出处
《计算机与数字工程》
2021年第2期305-309,共5页
Computer & Digital Engineering
关键词
交通流量预测
小波神经网络
加动量项
收敛速度
预测精度
traffic flow prediction
wavelet neural network
adding momentum term
convergence rate
prediction accuracy
