Short-term Prediction About Travel Time Of Interrupted Flow Based On Signal Cycle

It is essential that people can travel for their convenience under the guidance and the control of the Intelligent Traffic System(ITS).The links divided by the signal intersections are the constituent elements of the urban roads.Meanwhile,they compose the basic data sources and console units for the ITS.The travel time of each link shows the condition of traffic congestion and at the same time reflects the degree of the traffic congestion.It plays an important role in transportation planning and traffic management.The travellers,the traffic managers or the researchers often make their judgements about the traffic condition or make their choice for the travelling according to the travel time.In this paper,the writer intends to describe the traffic flow of the signal intersections of the urban roads as interrupted flow,and make short-term prediction about travel time based on the signal cycle.The main contents and the achievements of this paper are as below:First of all,the writer intends to make analysis of the characteristics of the traffic flow and divided the links into intersections segments and link segments accordingly.As the uninterrupted flow which is not affected by the signal cycle occurs,the writer uses the ARIMA algorithm to build model to estimate the travel time of the link segments.Based on the analysis of matching GPS data with map and of interpolation of location and time,the results shows that the travel time of link segments can be estimated accurately with the ARIMA algorithm.According to the red and yellow lights or the green lights in the signal cycle when a vehicle arrives at the intersection,the writer makes a further division of the travel time of the intersection segments.Thus,the writer puts forward the queuing model based on the improved Markov chain queuing theory and the traffic flow fluctuation theory.The writer builds the estimation model for the travel time of the intersection segments according to the time matching scheme of the signal cycle.It is illustrated that the model can better estimate the travel time of the intersection segments based on example analysis.Based on the above research and the spline function neural network model,the writer,considering and analyzing the key factors that affect the travel time,chooses the travel time of the link segments,the travel time of the intersection segments,signal cycle,the valid green light time and green time ratio as the key factors that affect the travel time and build the travel time prediction model.Compared with BP neural network,RBF neural network and Kalman filter,it is well illustrated that the spline function neural network model can better forecast the travel time of the experimental link,which are based on the travel time of the link segments,the travel time of the intersection segments and signal cycle.