Study On Multimodal Traffic Assignment Problem With Multiple Connections

Assignment theory is used to predict the distribution of travel demand on urban transport network and plays a crucial role in transportation planning and management as well as in policy evaluation.With the establishment of a transportation network consisting of multiple modes in cities,the assignment theory for a single mode can hardly be applied to the multimodal network.At the same time,urban problems such as unbalanced urban transport supply,low public transport sharing rate and inconvenient transfers between travel modes still exist.Therefore,the multimodal assignment theory studied in this thesis will aims to provide a basic theoretical reference for the design and optimization of urban multimodal transport network and the evaluation of transportation policies.The main content of this thesis consists of the following four parts:(1)Chapter 2 proposed a method for modeling the multimodal urban transport network.First,the necessity of considering common bus line problem when constructing a multimodal transport network is demonstrated by counting the ratio of overlapping lines in Shenzhen bus network.Then,in order to understand the impact of two network representations,route seciton and hyperpath,on the extended network scales,both are tested on six bus networks of different scales.The results shown that route seciton representation is more suitable for smallscale networks,and hyperpath representation is more advantageous in medium and large-scale networks.Finally,the existing method of modeling the multimodal network is improved by proposing two coupling method between transit network and road network,while a matching algorithm is used to ensure that stop(station)are mathced to a correct road link.Furthermore,the node contraction is used to create a fixed joint network.Compared with using transfer arcs to directly connect stops within 500 meters,a numerical experiment on Shenzhen transit network confirmed that this method can reduce the transfer arcs by about 85%.(2)Chapter 3 proposed several algorithms for solving the multimodal shortest vialbe hyperpath problem.First,multimodal viable hyperpath is defined,the constraints of the sequence of used modes are modeled by a deterministic finite automata(DFA).Then,it is proved that when searching for a shortest hyerpath on a multimodal network,the optimal attractive set needs to be determined for each label update of a bus stop to ensure the optimal solution.The label correctting(LC)algorithm for solving the multimodal shortest vialbe hyperpath problem is improved,and label setting(LS)and A*algorithms are proposed as well as the three algorithms respectively combined with multi-queue(MQ)to form MQLC,MQLS and MQA*algorithms.In order to test the computational efficiency of the six algorithms when respectively using Bellman condition,basic dominace rule and state-based dominance rule to update labels,numerical experiments were conducted on a large-scale network with 77787 nodes and 211825 arcs.The results shown that,compared with the LC algorithm,the LS and A*algorithms have significant advantages in computing speed,and MQA*has the fastest computing speed among the 6 algorithms.Regarding the dominance rules,the time-consuming order of three dominance rules is:Bellman condition>Basic dominance rule>State-based dominance rule.The computational process can be accelerated when the number of check labels is significantly reduced by the multi-queue.(3)Chapter 4 established a multimodal equilibrium assignment model.According to the characteristics of travel modes,the generalized time cost function on each mode is defined by considering two factors,trip time and money spent.Then,the hyperpath-based transit equilibrium assignment model is extended to multimodal transport network,and a multimodal variational inequality model equivalent to the user equilibrium assignment problem is formulated.(4)Chapter 5 designed an algorithm for solving the equilibrium assignment problem in multimodal transport network.Referring to the gradient projection(GP)for solving the hyperpath-based transit assignment problem,a algorithm for solving the multimodal equilibrium assignment problem is designed by replacing the network model and shortest path algorithm with multimodal transport network and multimodal shortest vialbe hyperpath algorithm.In addition,an inner looping strategy is also introduced to combine GP algorithm to speed up the solution efficiency,and the GP algorithm with inner looping is called the iGP algorithm.The algorithms were tested on small,medium and large networks,respectively.Numerical experiment shown that both GP and iGP algorithms can effectively solve the multimodal assignment problem,and the inner looping strategy can significantly improves the efficiency and stability of the algorithm.Finally,the impacts of two parameters,transfers and value of time,on path choice and flow distribution are analyzed on a small network.In summary,considering the multiple connections of different transportation modes,this thesis studies the multimodal equilibrium assignment model and algorithm for large scale networks.This research can provide theoretical reference for the scientific development and management of integrated urban transport systems.