Reliability Study of Urban Expressway Performance Based on Artificial Neural Network

The increase of traffic demand leads to the instability of urban expressways,which requires a probability measure parameter such as reliability to evaluate the dynamic performance of expressways.First,the concept of expressway performance reliability was defined and the influence factors such as saturation,density of onramp and off-ramp,fluctuation of traffic demand,traffic condition of corresponding side roads were analyzed.And then the construction of the functional relations between reliability and its factors by neural network was proposed.On the basis of different input variables normalization,the reliability model of BP neural network was built.Finally,the proposed model was testified by a numerical example,and the result indicated that the maximum relative error between factual output and target output was less than 3.5%.