| The dynamic emission estimation of moving sources such as vehicles has always been the bottleneck of air quality simulation in China.Air quality simulation based on the dynamic emission of various pollution sources is the important reference for decision-making in air pollution control.At present,the fixed sources such as industrial sources have realized online real-time monitoring,which can obtain continuous dynamic emission data.However,the moving sources such as vehicles are mostly annual static total emission estimation,which is difficult to support the real-time dynamic estimation and air quality simulation analysis of air pollutants.The lack of effective estimation methods of dynamic flows on whole road network is one of the important reasons for the difficulty to estimate dynamic vehicle emission.The dynamic vehicle emission is estimated based on dynamic traffic flow,link length and emission factor.The estimation of single vehicle emission factor has been deeply studied by a large number of researchers,there exists an effective method of the estimation of single vehicle emission factor.However,due to the limitation of monitoring technology,it is difficult to obtain dynamic traffic flow of the whole road network.Besides,there is no effective method for the estimation of dynamic traffic flow of the whole road network.These two reasons lead to the difficulty of estimating the dynamic vehicle emissions.Thus,it is essential to study the method of dynamic flow estimation of the whole road network.At present,the technology of collecting dynamic speed of the whole road network is mature.Based on the traffic fundamental diagrams,the flow can be estimated through the speed data,which is an important method to obtain the traffic flow of the whole road network.The research about the traffic flow estimation based on traffic fundamental diagrams have been carried out earlier.However,for the traffic control,the traffic flow for short time granularity(5 minutes)can not be accurately estimated based on traffic fundamental diagrams,which leads to its limited application in practice.For the dynamic estimation vehicle emissions on the whole road network,the hourly traffic flow can achieve the dynamic emission estimation,at the same time,the estimation of the hourly traffic flow based on the traffic fundamental diagrams is of high accuracy.Therefore,the method of traffic flow estimation for the whole road network was developed based on the traffic fundamental diagrams in this dissertation.The characteristics of road traffic is the description of the characteristics of the travelers’ trips in the road network.Influenced by trip purpose and road function,the traffic shows different patterns,for example,the work day traffic pattern based on commuting trips and weekend/holiday traffic pattern based on entertaining trips have different patterns.The traffic pattern at radiation roads was different with that at ring roads,which have different single and double peak morphology.The traffic patterns can be clustered into limited types due to the regularities of trips and the classifications of road types.Due to the significant difference between different traffic patterns,the traffic flow were predicted in different patterns,which can improve the accuracy of flow estimation.Therefore,it is necessary to study the clustering and identification of traffic patterns.In this context,based on multi-source field observed speed,flow and emission data,the clustering and identification of traffic patterns were researched,the influencing factors of the traffic fundamental diagrams were quantitatively analyzed,the traffic fundamental diagrams at different road types for different traffic patterns and periods were developed,the traffic dynamic flows on the whole road network were estimated,on this basis,a case study was designed to estimate the dynamic vehicle emissions on the whole road network.The findings in this dissertation were concluded as follows:(1)A traffic patterns clustering method was developed based on the direct index of time-varying speed by SOM neural network.It can achieve efficient clustering of urban traffic flows.Eight traffic flow patterns were constructed,which can effectively depict more than 90% of urban road traffic flow patterns,such as Monday to Thursday working day,Friday working day,Saturday,Sunday,rainy day,holiday,outstanding evening peak and outstanding morning peak patterns.(2)The rapid recognition methods of traffic patterns based on the speed indexes in different periods were developed.The pattern recognition based on 24-hour,morning-evening peak,0:00-12:00 and morning peak speed indexes all have high accuracy,the average correct recognition rates under these indexes are 94.87%,93.64%,82.45% and 80.96%,respectively.DBN algorithm can effectively recognize large-scale traffic flow time series,with the correct recognition rate of 93.02%,90.98% and 82.45%for 24-hour,morning-evening peak and 0:00-12:00 speed indexes respectively.The accuracy of traffic flow pattern recognition is improved by GA-BP and SAGA-BP,and the recognition accuracy of 24-hour speed indexes is improved by 7.38% and 7.96%respectively compared with BP algorithm.(3)The estimation method of dynamic flows on whole road network was established for different road types under different traffic patterns and different periods based on traffic fundamental diagrams.Through the quantitative analysis of the influencing factors of traffic flow-speed-density fundamental diagrams on specific road,it is found that in addition to the proportion of large vehicles and weather conditions that have been identified by existing studies,the driving behavior differences caused by travel purposes in different scenes also significantly affect the traffic fundamental diagrams,such as morning or evening peak,working day or holiday.The morning peak capacity of the same road is 3.47% higher than the evening peak,and the holiday capacity is 4.73% higher than the working day.The accuracy of traffic flow model considering road grade,morning and evening peak,working day and holiday is improved by 6.51%.(4)The case study of dynamic emission measurement based on dynamic flow on the whole road network was developed,which verifies the reliability of the method proposed in this paper.The implementation effect of energy conservation and emission reduction policies can be quickly evaluated. |
PDF Full Text Request