Car-following model considering jerk-constrained acceleration stochastic process for emission estimation

Abstract

The continuity of acceleration changes is often overlooked by existing car-following models, leading to a limitation in capturing realistic driving dynamics for emission estimation, which are essential for the application in microscopic traffic evaluations. This paper investigated and modeled the jerk-constrained acceleration stochastic process using the Markov model. A new car-following model considering the acceleration stochastic process was proposed, which incorporated two modes of unconscious following and active acceleration approaching. Additionally, a bi-objective model calibration framework was introduced to balance the trade-off between traffic-related performance and emission estimation performance. Numerical simulation was conducted to compare the performance of the new model with the conventional Wiedemann model. Results demonstrated that the proposed model provides more realistic vehicle dynamics and accurate emission estimations. Specifically, compared to the Wiedemann model, the new model reduced the root-mean-square error (RMSE) of spacing headway by 0.73 m and the RMSE of vehicle-specific power (VSP) distribution by 11.57%.