Quality of service is a method to evaluate the signalized intersections and level of service (LOS) is a hierarchy for assessing the quality of service. Many parameters may influence the quality of service such as delay. It is deserved to predict quality of service at signalized intersections since the traffic flow at this position is interrupted and capacity or saturation degree would vary from time to time. An accurate predicting quality of service at intersections would help to better management and control. This paper introduces neural network with SoftMax classification and Long Short-term Memory (LSTM) model which are proposed to predict LOS and delay respectively. A case is studied to validate the practicability and accuracy of the quality of service prediction model. The consequence indicates that the model is effective to predict LOS and delay. However, it needs more improvements to increase the accuracy.

CICTP