Study On Forecasting Method Of Urban Arterial Road Traffic Congestion

The rapid development of the city has attracted more and more traffic flow, led to traffic congestion becoming bottlenecks restricting the development of the city, the problem many cities in the world facing and waiting to be solved. All places, on the one hand, ease the traffic problem by increasing the road infrastructure construction, on the other hand, combining with the intelligent traffic system to governance congestion has become an important method of grasping the corrosion, ensuring the unclogging roads, improving people’s livelihood. As an important part of intelligent transportation system, congestion prediction is an effective measure to alleviate urban traffic congestion.Based on the analysis of the concept, classification and quantification standard of traffic congestion, in view of the current shortage of urban road traffic state according to the prediction of traffic flow parameters, this paper selects three aspects indexes, traffic flow, drivers’ feelings and microscopic driving behavior, predict the selected indexes respectively using the autoregressive integrated moving average(ARIMA), two exponential smoothing, and BP neural network, establish a comprehensive prediction model based on the multi model fusion technique, and forecast the indexes, test the prediction accuracy, and verify the feasibility and validity of the comprehensive prediction method.Collecting mass historical data of urban trunk road, the paper uses the analytic hierarchy process and the K-means algorithm to establish a comprehensive traffic state classification algorithm, classify the actual traffic state, and mark them four grades such as smooth, normal, mild congestion, severe congestion, identify the traffic state using the forecasting data, verify the rationality and reliability of the discriminant method. Response to the growing problem of urban traffic congestion, the paper discusses the ease urban traffic congestion ideas from the macro policy and based on forecast traffic state.This paper attempts to establish the comprehensive prediction and discriminant method of city road traffic congestion from the traffic flow parameters, drivers’ feelings and microscopic driving behavior, has some reference value to the research of city road traffic congestion prediction.