基于BPNN-TD算法的城市轨道交通线网规模预测方法

Prediction Method of Urban Rail Transit Network Scale Based on BPNN-TD Algorithm

科学预测城市轨道交通线网规模,对于轨道线网的规划建设与城市布局发展具有重要意义。基于轨道交通线网规模及其影响指标数据,综合两种模型优势,对城市轨道交通线网规模进行有效预测。首先,从政策、经济、城市规模、出行需求4个方面,简要分析城市轨道交通线网规模的影响因素,并利用相关性分析法,提取GDP、第三产业值、人口规模、建设用地规模、日均客运量等5个模型输入指标。其次,构建基于BP神经网络模型(BPNN)的城市轨道交通线网规模预测方法,在求解熵权向量的基础上,结合交通需求法(TD)调整预测结果。最后,以广州市轨道交通线网规模为例,以误差最小为目标,对模型的隐含层数、神经元数、激活函数等进行优化。研究结果得出:2023年广州市轨道交通线网规模预测值为745.2 km,低于实际规划值5.9%,表明广州市轨道交通线网规模的发展规划仍存在调整空间,研究有助于在大数据背景下为城市轨道交通线网的规划设计提供理论支撑。

The scientific prediction of the the future scale of urban rail transit networks is crucial to the planning,construction of urban rail transit networks,and urban layout development.This study investigates the effective prediction of the urban rail network scale based on the urban rail network scale data and impact indicators by combining the advantages of two models.First,this work briefly analyzes the factors influencing the scale of the urban rail network from the aspects of policy,economy,city scale,and travel demand.Then,the gross domestic product,tertiary industry value,population scale,construction land scale,and daily passenger volume are extracted as model input indicators based on the correlation analysis method.Second,the prediction model of the urban rail network scale is formulated based on the back propagation neural network model,and the traffic demand method is employed to adjust the prediction results by solving the entropy weight vector.Finally,considering the scale of the Guangzhou railway network as an example and the minimum error as the objective to optimize the number of hidden layers and neurons,the activation functions of the proposed model are identified.The results show that the scale value of the Guangzhou railway network predicted for 2023 is 745.2 km,which is 5.9%lower than the actual planning value.This indicates that the scale of the development planning of the Guangzhou railway network can still be improved.This study provides theoretical support to the planning and optimization of urban rail transit networks.