Optimization Of Wiedemann And Fritzsche Car-Following Models For Emission Estimation

As Chinese urbanization accelerates and vehicle population booms, vehicle exhaust pollution becomes one of the main problems that affect our daily life. Microscopic traffic simulation model has served as the mothed to link the evaluation of urban traffic flow with estimation of vehicle emissions, and its evaluation function on urban traffic emissions are gaining more and more applications and researches. However, because of their inner simulation-mechanism, car-following model, which is the core module of microscopic traffic simulation models, could not meet the satisfactory accuracy for vehicle emissions. Thus, based on VSP distribution, which proves to be the best explanatory variable for vechicle emission calculatioin, this paper focuses on the methods for imrpoving the accuracy of Wiedemann model and Fritzsche model on emission estimation, and verifies optimization results in the end.Firstly, this paper analyzes physiological-psychological model and investigates its inner following logics and simulation mechnisum. Besiedes, the study also analyzes the worldwide researches on microscopic emission model and the optimizations of car-following models for emission estimations.Secondly, this paper gives a detailed method on data collection by using portable GPS devices and the design of simulation method. Massive data of actual driving behavior and typical following-driving behavior on the expressways in Beijing are collected by portable GPS. Based on the collected field driving and car following data, and the regularity that the distribution characteristic of driving behavior keeps consistant on the same road types, this paper designed the numerical simulation method in Matlab for Wiedemann and Fritzsche model.On this basis, this paper conducts sensitivity analysis on the parameters of SDV, CLDV and SDX for speed and headway regimes in Wiedemann model based on the time fractions of different following states simulated by Wiedemann and Fritzsche model. According to the results of sensitivity analysis, the parameter of SDX for headway regime is choosen for optimization, and the improved SDX model for Wiedemann is proposed. Besides, based on the massive data of actual driving behavior, this paper analyzes the relationship model between maximum acceleration and instaneous speed, and replaces the traditional expressions of maximum acceleration in Wiedemann and Fritzsche models. After analysing the overall effects of optimization, the study finds that the average RMSE of VSP distribution in different speed bins from Wiedemann model is reduced significalty from0.565to0.102. By comparing with the ogrinal model, the average RMSE of VSP distribution in different speed bins from the optimized Fritzsche model is reduced by57.23%.In the end, on the basis of the collected car-following data, this study designs a stability control algorithm for car-following models, and verifies the model stability of optimized Wiedemann and Fritzsche model from the aspects of headway, reasonability of simulated acceleration, and RMSE between simulated and field speed. Results show that the optimized Wiedemann and Fritzsche models still demonstrate good stability.