Abstract

It is known that failure dependencies in complex systems may cause cascading failures among components and dramatically reduce system reliability. Urban road network is a typical example where failure dependencies often exist between roads, leading to failure (congestion) propagations. Thus, identifying strong failure dependencies and their clusters in road networks can help city managers to relieve congestions. However, previous road network reliability studies seldom considered to uncover existing dependency clusters. Here, we propose a causal inference approach to detect dependency relationships among road sections based on information theory using real-time velocity data in Beijing. On the obtained traffic causal network, percolation is used to identify smaller dependency clusters, which will merge into a giant dependency cluster at the percolation criticality. Critical dependency links contributing to the formation of the giant dependency cluster have been also detected. Moreover, the power-law distribution of finite-size dependency cluster sizes at criticality has been investigated, where the estimated exponent can be related to the log-normal degree distribution of the causal network. Finally, we illustrate how the identified dependency clusters impact the road network reliability. Our findings can help to uncover potential cascading failure patterns in road networks, and promote reliability management of transportation systems.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call