基于复杂网络与机器学习理论的城市公交网络客流模型研究

客流模型常用于估算或预测城市公交网络的客流分布，辅助其设计、优化与调度，对提升公交服务性能、缓解城市交通拥堵有积极作用。现实中，乘客在城市公交网络中出行往往有多条可选路径，传统客流模型研究较少考虑乘客的路径选择行为。本项目开展基于复杂网络与机器学习理论的城市公交网络客流模型研究，内容包括：建立可对城市公交网络进行定量表示的网络描述模型，在此之上建立乘客出行候选路径算法；基于公交大数据开展城市公交网络乘客出行路径选择行为模式研究，使用机器学习理论建立估算各候选路径被乘客选用概率的乘客出行路径选择模型；基于乘客出行候选路径算法、路径选择模型，使用复杂网络配流理论建立估算给定OD需求分布作用下各站点与线路段上通过客流的城市公交网络客流模型。本项目综合考虑了城市公交网络的网络与乘客出行路径选择行为因素，研究目标为建立高效准确的城市公交网络客流模型，形成实用的客流仿真方法，具有重要的理论与应用价值。

Passenger flow models are often used to estimate or predict the distribution of passenger flow in urban public transport networks, assist in the design, optimization and scheduling, and have a positive effect on improving public transport service performance and alleviating urban traffic congestion. In reality, passengers often have multiple alternative paths when using public transport networks to travel. Traditinoal studies seldom consider the path choice behavior of passengers when modeling passenger flow. This project carries out research on urban public transport network passenger flow model based on complex network and machine learning theory. The content includes: establish a network description model that can quantitatively represent urban public transport networks, and establish a passenger travel candidate path algorithm on top of it; carry out research on passenger travel path selection behavior patterns of urban public transport networks based on public transport big data, and use machine learning theory to establish a passenger travel path selection model that estimates the probability of each candidate path being selected by passengers; based on the passenger travel candidate path algorithm and path selection model, use the complex network traffic assignment theory to establish an urban public transport network passenger flow model that estimates the passenger flow of each station and line segment under the action of a given OD demand distribution. This project comprehensively considers the network and passenger travel path selection behavior factors of urban public transport networks. The research goal is to establish an efficient and accurate urban public transport network passenger flow model, and form a practical passenger flow simulation method. It has important theoretical and application value.