摘要
为提高历史文化街区的交通规划手段,利用出行者特征,出行特征、交通特征与交通工具特征等属性作为BP(Back-Propagation)神经网络输入层数据,建立一套基于传统贝叶斯正则化算法(Bayesian Regularization)的出行方式选择模型。通过对珠海市6个历史文化街区的实证分析,发现基于传统BP神经网络的样本训练后预测值准确率高,各地区交通分担率对比的总体平均误差约为1.02%,与实地调查的实测值基本吻合。研究结果表明基于传统BP神经网络的交通预测模型在历史街区中的应用有效,能为历史街区交通规划作出一定理论支持。
In order to improve the methods of traffic planning in historical and cultural blocks,the characteristics of travelers,travel features,traffic characteristics,and vehicle attributes are used as the input data of the Back-Propagation(BP)neural network to establish a travel mode choice model based on the traditional Bayesian Regularization algorithm.Through empirical analysis of six historical and cultural blocks in Zhuhai city,the study found that the accuracy of the predicted values after sample training based on the traditional BP neural network is high,and the overall average error of the traffic share rate comparison in various regions is about 1.02%,which basically matches the actual measurement values from field surveys.The research results show that the traffic prediction model based on the traditional BP neural network is effective in historical blocks and can provide certain theoretical support for the traffic planning of historical blocks.
作者
莫育强
MO Yuqiang(Beijing Institute of Technology,Zhuhai,Guangdong Zhuhai 519088 China;Faculty of innovation and design,City University of Macao,Macao 999078 China)
