摘要
针对短时交通流量预测的难题,在传统贝叶斯组合模型进行改善的基础上,提出一种改进型贝叶斯组合模型.该模型只根据各基本预测模型当前时刻之前几个交通流量的预测表现,通过提出的分配算法实时更新组合模型中各个基本预测模型的权重,从而改善了传统贝叶斯组合模型权重计算迭代步长过长的缺陷,提高了贝叶斯组合模型对各个基本预测模型预测精度的灵敏性.通过对实地的交通流量的预测发现,基于改进型贝叶斯组合模型的预测精度不仅优于单一的预测方法,而且也优于传统的贝叶斯组合模型,从而证明了改进型贝叶斯组合模型有效提高预测的可靠性和具有一定的实用性.
To solve the problem of short-term traffic flow prediction,a new method called improved Bayesian combined model is put forward based on the improvement of the traditional Bayesian combined model.This method can update each basic prediction models＇ weights only by its performance in the past several times.Thus the defect of over iteration steps for calculating the weights in traditional Bayesian combined model can be corrected.The improved Bayesian combined model is more sensitive to the accuracy of each basic prediction model.According to the performance of the practical traffic data prediction,the results of improved Bayesian combined model are not only better than the single prediction model,but also better than traditional Bayesian combined model.Consequently,it is regarded that the improved Bayesian combined model increases the credibility of the prediction,and is applicable for the real condition.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2012年第1期162-167,共6页
Journal of Southeast University：Natural Science Edition
基金
"十一五"国家科技支撑计划资助项目(2006BAJ18B03)
关键词
贝叶斯组合模型
交通流
小波分析
ARIMA算法
BP神经网络
Bayesian combined model
traffic flow
wavelet analysis
autoregressive integrated moving average algorithm
back propagation neural network
