Shortest path problem in stochastic time-varying transportation networks considering risk measurement

Stochastic, time-varying networks are proved a better tool in modeling transportation networks. In real transportation networks, routing policies often rely on shortest path, and finding shortest path is also proved a NP problem in STV networks. This paper presents a novel shortest path model based on expectation disutility function. The function portrays its risk measurement. In order to decrease the complexity of computation, the paper devises an approximation method in describing the path travel times. The method computes the travel time according to its distribution of arriving time of each node in the path. Compared to known methods, the approximation method prevents uncertain nesting computation coming from recursive formula, which is time-consuming in the face of large scale networks. Theoretical analysis and numerical example prove its efficiency and feasibility.