Research On Urban Road Travel Time Estimation Based On Automatic Number Plate Recognition Data

Real-time,accurate and reliable travel time information can directly reflect the traffic state of urban road network,and is an important support for intelligent travel information system and intelligent traffic management system in intelligent transportation system.As an open road network,the vehicle will be affected by multiple factors such as random delay,signal control scheme,random arrival and departure and so on,which results in obvious interrupted traffic flow characteristics.As a result,the travel time estimation of urban road is more uncertain than that of continuous flow facilities such as highway and expressways.Therefore,this paper fully considers the effects of signal and builds a new set of urban road travel time estimation method based on license plate recognition data.The specific research is as follows:(1)A set of systematic travel time data acquisition and preprocessing method is proposed for the license plate recognition data,and the data scale,license plate recognition accuracy,sample matching rate in the study area are analyzed in detail,the results show that the data source has a high reliability and stability.(2)Considering the influence of signal control on travel time,a travel time estimation strategy based on the characteristics of intermittent traffic flow is proposed.Urban road travel time is divided into non-stop component,signal control delay component and activity component.Activity vehicle data is defined as outliers,respectively build non-stop component estimation model based on gaussian mixture distribution model,signal control delay component estimation model based on the delay pattern model.A more accurate threshold window estimation of cycle level is realized,and the travel time adaptive filter is used to clean and estimate the travel time,nd the travel time is filtered by the real-time threshold window result.(3)The method is applied to the study section to complete the adaptive filtering of lane level travel time cleaning.The estimated results of this algorithm are compared with those of the median average filter algorithm and Dion low-pass filter algorithm,and mainly from the filter point,threshold window,mean estimation results and other aspects of qualitative and quantitative analysis and evaluation.The results show that this algorithm can not only provide a more accurate,reliable and stable threshold window,but also retain the signal control delay data while effectively suppressing the high-frequency active noise data,which can better reflect the real delay state of urban roads and the change trend of travel time.