A Research Of Short-term Prediction And Control Technology Of Urban Traffic

With the rapid development of economy in our country,urban traffic travel demand is growing concurrently.At the same time,because of the irrationality of the existing path planning and the tardiness of new-build road,urban traffic congestion status gets more serious.At present,most of urban traffic lights using off-line single-point timing control strategy,cannot be adjusted according to changes in traffic flow conditions automatically,which makes by optimizing the control strategy of the single intersection may ease the traffic congestion.Therefore,directing at urban single intersection,it`s seems particularly import to study and put forward a method that can adaptive adjust and optimize signal control strategy according to the intersection traffic flow conditions.On this basis,design and implementation of an intersection traffic signal control system have very realistic significance to alleviate urban traffic congestion problem.In this thesis,the main content and innovation points are as follows:(1)This thesis puts forward a traffic flow short-term prediction model based on fuzzy time series.Focus on the inadequate of classic fuzzy time series in the domain division and establishment of fuzzy relation,this model utilize fuzzy C-average clustering method,weighted method based on importance of FLR,membership vector weighting method to optimize classic model,and improve the prediction accuracy.(2)This thesis puts forward an intersection traffic flow state division method.This method uses the historical traffic flow data as the training sample,and establishes the traffic flow state clustering model by identifying and defining the traffic flow state of each lane and phase of the intersection,which makes the state division more reasonable.(3)This thesis puts forward an adaptive control method for traffic lights.This method can judge the traffic flow state of the intersection in the next period of time based on the predicted value by the traffic flow short-term prediction model,so it can adjust traffic lights control strategy dynamically.In addition,within the forecast cycle,a micro-correction can be made by the traffic data collected in real-time data.Based on the above innovation points,this thesis designs and implements an online micro-traffic simulation platform and a traffic signal control system.Experimental results suggest that the short-term traffic flow prediction model,based on fuzzy time series,can make a more accurate prediction on more than 5 minutes traffic flow.Moreover,compared with the traditional single point timing control method,adaptive control method,raised in this thesis,can effectively reduce the vehicle waiting time and alleviate the traffic congestion status.