Quantitative Analysis Of Road Traffic Energy And Emissions Based On Traffic Flow Model Transformation Under Data-Driven Conditions

The increase in energy consumption and greenhouse gas emissions due to road traffic is one of the current widespread concerns worldwide.In order to effectively quantify the energy consumption and emission of regional traffic,macroscopic and microscopic traffic flow models are generally used to couple with energy consumption and emission models.However,due to the difficulty of obtaining accurate traffic data and the existence of parameter approximation errors when coupling different levels of models,the energy and emission estimation results are not objective and accurate enough.To this end,for the problem of energy and emission estimation error when coupling different granularity traffic flow models and energy consumption and emission models,this paper proposes a comparative analysis method of road traffic energy consumption and emission by coupling macroscopic and microscopic models.Taking the macroscopic California Pe MS data and microscopic NGSIM data as the research objects,we reconstruct the vehicle trajectory by multi-source data fusion method for Pe MS data.And we perform macroscopic traffic state aggregation for NGSIM data based on Principal Component Analysis(PCA)and K-Means algorithm.By combining the macroscopic California EMFAC and microscopic MOVES energy and emission models to construct a comparative energy emission validation framework with temporal and spatial consistency.And we crossvalidate the energy and emission estimation errors under the coupling of different granularity traffic flow models and energy consumption and emission models.The specific research work of this paper is as follows:(1)A spatio-temporal consistent energy consumption and emission comparison validation framework is proposed.The Pe MS data and NGSIM data of the same road section at the same time in California are selected as the object of the study.The Pe MS data are reconstructed into vehicle trajectories,and the combination of "macroscopic traffic flow + microscopic energy consumption and emission" is established with MOVES model.NGSIM data are aggregated with macroscopic traffic states,and the combination of "microscopic traffic flow + macroscopic energy and emission" is established with EMFAC model.The estimation results of "macroscopic traffic data Pe MS + macroscopic energy and emission model EMFAC" and "microscopic traffic data NGSIM + microscopic energy and emission model MOVES" are used as the evaluation benchmark to cross-validate the estimation errors under different combinations of granularity models.(2)For the problem of microscopic traffic data characterization under known macroscopic traffic data,a vehicle trajectory reconstruction method based on multi-source data fusion is proposed.By fusing fixed detector and mobile probe vehicle data,the traffic speed evolution of the road spatio-temporal network is estimated by a nonparametric kernel smoothing method.Under the sparse probe vehicle trajectory constraint,the complete vehicle trajectory is reconstructed based on the variational theory of three-dimensional(3D)kinematic wave theory and shortest path method.Experimentally,the accuracy of trajectory reconstruction is compared and analyzed with different cell sizes and probe vehicle proportions as variables to determine the best combination of parameters.Meanwhile,by comparing with the four-corner interpolation method and 3D kinematic wave theory,it is verified that the trajectory reconstruction accuracy of the method in this paper is high.(3)For the problem of macroscopic traffic data characterization under known microscopic traffic data,a traffic state estimation method based on PCA and K-Means is proposed.Based on the PCA method,the high-dimensional parameter space related to vehicle speed is downscaled into a low-dimensional space.Based on the K-Means method,the downscaled data are clustered into different clusters,and the average speed of road traffic,traffic flow,and traffic density are calculated by weighting the size of each cluster.The experiment verifies the effectiveness of the traffic state estimation method in this paper by comparing with DBSCAN clustering method,time mean speed and space mean speed,and NGSIM traffic data.The experimental results show that 1)The trajectory reconstruction method and the traffic state estimation method proposed in this paper has a good estimation effect,and the fuel consumption emission results are closest to the real micro-level and macro-level data.2)Crossvalidation of different macroscopic and microscopic models shows that the more detailed characterization of regional traffic and the higher accuracy of the energy consumption and emission model,the more accurate the calculated results are.