Dynamic Estimation And Reliability Calculation Method Of Travel Time Of Important Urban Traffic Passages

Travel time is a key indicator to reflect the operating efficiency of road network.It can well evaluate the unblocked condition of the road and play an important role in traffic control and travel planning guidance.When traveling during the peak period,the only option is to leave in advance to arrive at the destination on time.Therefore,it is of great significance to estimate the travel time and calculate the reliability of the travel time for the planning and guidance of travelers.Based on this,this paper takes part of the road section in Chongqing as the experimental section,and uses the monitoring data of python crawling section and online data of Aude map to study the dynamic estimation of travel time and the calculation method of reliability.Major work and results accomplished include:Firstly,the relationship between traffic demand and travel time under different traffic states(i.e.non-crowded state and crowded state),as well as the influence of some accidental events on travel time are analyzed,so as to determine that the two main influencing factors of travel time are mainly periodic and random fluctuation of traffic demand.The rationality of these two influencing factors is verified by the travel time distribution characteristics on the actual path.Then the basic concept of travel time reliability and several common travel time distribution functions as well as the indexes used to evaluate the travel time reliability are introduced,which is the basis for the prediction of travel time distribution and reliability calculation in the following part.Secondly,according to traffic demand and random fluctuations,a computational model of road section travel time is established,and a probability distribution model of road section formation time is established by considering the variation law of road section travel time under different traffic states,and the parameter estimation is carried out by using the maximum likelihood method.Based on the dynamic characteristics of traffic demand and the correlation between sections,a path travel time estimation model is established.The gaussian distribution,lognormal distribution and gamma distribution were used to verify the fitting distribution of the travel time of the actual section,and the fitting effect was analyzed by calculating the P-value,SSE and R-square values.The results showed that the normal distribution was indeed good for the fitting of the travel time.Finally,combining the travel time distribution function and travel time reliability threshold of section and path,the travel time reliability calculation models of section and path are established respectively.Based on LSTM neural network prediction model,BP neural network prediction model and AMRA time series prediction model,a comprehensive prediction model of predicted travel time is established to compare the predicted travel time with the calculated reliability results under the specified threshold.The results show that most of the predicted travel time is less than the specified travel time threshold,which verifies the accuracy of the calculated travel time reliability under the specified travel time threshold.