Freeway Network Vulnerability Analysis Based On Multi-source Data

Freeway network, as a key part of the transportation industry, is the material foundation of the regional linkages and social activities. However, all kinds of emergencies tend to impact on the performance of the freeway network, that is, freeway network possesses vulnerability to some extent. Vulnerability would cause performance degradation of freeway and even different degrees of injuries or damage. With the development of economy and the continuous improvement of freeway network, research on freeway network vulnerability has been paid more and more attention from managers and researchers.Multi-source and massive traffic data, including traffic flow and traffic accidents data, can be obtained by traffic managers due to the development of traffic data collection devices. It has been a cutting-edge topic for freeway network researchers to carry out vulnerability assessment of freeway network using data mining technology based on multi-source and massive traffic data, aiming to propose reasonable optimum proposals and improvement measures. Therefor, this pape aims to propose a method, which is based on the objective data and the game theory, to assess the freeway network vulnerability and identify the critical links so as to develop specific and workable proposals for freeway network.Factors which affect the vulnerability of freeway network are analyzed from three levels: link-level, node-level and flow-level. The impact mechanism is analyzed from the perspective of internal and external factors. These factors include network attributes, threat attributes, flow attributes, and neighborhood attributes. Then, correspondingly, data needs, as well as the data sources, are elaborately listed according to the influence factors. Thus a relationship between affecting factors of network vulnerability and multi-source traffic data is established. And the necessity and application value of multi-source traffic data on freeway network vulnerability are obvious.Based on the above analysis and the data collected by freeway toll collection system, which is large and easily accessible, fused with the meteorological data and traffic accidents data, a novel algorithm of calculating the section traffic flow is proposed by analyzing the impact on traffic flow of weather and accidents; and according to the different inflow between two cells, time series model is constructed to reflect traffic flow state. Then minor accidents on freeways, which are often unrecorded, can be identified using temporal data mining. Then based on statistical theory, SPF is used to identify traffic accident-prone locations. Then the results of section traffic flow calculation and accident black spots identification are the important input data of critical links identification and vulnerability assessment.In regard to the abstract model, the game theory is adopted in the paper. The network vulnerability is abstracted out as a destroyer and the traffic managers are the defender. They are two players in a game. The destroyer aims to destroy the network and decrease the network performance as much as possible, while the defender try its best to minimize the negative affects for users of the network destruction. Thus a game theory is embeded. In this game framework, the network vulnerability assessment and countermeasures are carried out synchronously.Meanwhile, to solve the game problem, two algorithms are proposed. The defender assigns the traffic based on the shortest path assignment and entropy function is used by the destroyer to ruin the network without the consideration of propagation of traffic congestion. Taking congestion into consideration, the defender assigns traffic through the Frank-Wolfe algorithm to achieve the user equilibrium condition while the destroyer calculates its failure strategy through the Interest function. The network vulnerability and critical links ranking, as well as the countermeasures, are the output when the game terminates.To verify the effectiveness and applicability of this method, this paper takes freeway network of Heilongjiang province as an example. Using the multi-source data(including the network-related data, freeway toll data, meteorological data and traffic accidents data) collected on freeways in Heilongjiang province, the game problem is solved by two proposed algorithms. Then through comparative analysis, the results are evaluated and related policy suggestions are given reasonably.