Research On Evaluation Of The Level Of Service Supplied By Intercity Freeway Basic Section Based On Short-term Traffic Flow Prediction

As a result of excess increasing of traffic in the economic developed region, traffic capacity of intercity freeway aggrandizes increasingly. Especially in a section of the road into and out of the city zone, the traffic ability is saturated, accidents happen frequently, and congestion is severe. The section has become the bottleneck of restricting city economy and sustainable development of traffic. In the situation of road facility is impossibly implemented in short time, optimizing and perfecting road traffic control and management by technique becomes an effective countermeasure. Evaluating the freeway level of service is one part of ITS, if accurately forecasted, the action will be leaded, traffic flow be equipoised, traffic jam be decreased. But LOS theory applied to freeway exists limitations, yet to be further improved. Therefore, researching formulation of LOS has a great theoretical and practical significance as to forming LOS determing method, and promoting the healthy development of highways.This paper summarizes traffic characters of the intercity freeway, takes Jingjintang freeway for example, educes correlation of parameters, and expatiates other parameters infection on traffic flow, considering flow, speed and occupancy in gray system predication. Meanwhile, there is in-depth analysis of tradition short—term traffic flow prediction models, we use genetic algorithms for optimizing BP artificial neural network. From the viewpoint of data fusion, the short—term compound forecasting is put forward to. Based on the hypothesis of traffic continuity, pressure formula is founded, forecasting model of intercity freeway LOS is built. Additionally, comparing of different methods comprehensively, we decide to choose BP-GA, GM, ARIMA and hybrid prediction model. Jingjintang freeway has been taken as an example to forecast traffic flow, and four results are compared. Results show that the hybrid model, which takes advantage of the unique strength of the three models can produce more accurate predictions than that of single model. Finally the paper predicts LOS based on hybrid result, comparing with tradition prediction, testing and verifying the practicability of LOS predicting model. Calculation results show that the method is reasonable and feasible and will provide the basis for freeway management and control in the future.