Analyzing Safe Spacing Threshold of Follower Vehicle of Reaction Point based on Behavioral Driver Using Artificial Neural Networks

Various reasons lead to traffic oscillation, one of which is the sudden drop in the speed of the leader driver. The stop and go traffic wave along with two parameters [τ,δ] is propagated toward the upstream based on the Newell s car following model. The follower vehicle drivers respond differently to the reception wave based on their intrinsic behavior characteristics, which it leads to the deviation of the follower driver s behavior from the ideal driver s trajectory, Newell. This article is classified the follower driver s behavioral patterns based on the asymmetric behavioral theory in the deceleration phase and the hysteresis phenomenon in the acceleration phase, and different behavioral pattern of the follower vehicle driver in the NGSIM trajectory data. By fixing the parameters τ, δ, the hypothesized direction of the Newell driver is identified and the degree of deviation of the follower driver s behavior from Newell driver s path is also determined. The follower driver responds differently to the reception deceleration wave based on any behavioral pattern, which leads to secure a safe spacing and to change behavior at the behavioral change point. Then, the neural network models are developed to analyze the effective parameters at the microscopic level on the safe spacing of the follower driver at the behavioral change point based on different behavioral patterns. The analysis results show that the most effective parameters on the follower driver s safe spacing at the behavioral change point are two independent parameters of the follower vehicle driver s speed at the wave reception point and the deceleration wave leading to congestion based on the over reaction-timid behavioral pattern, and the parameter of the deceleration wave leading to congestion based on the under reaction-timid and over reaction-aggressive behavioral patterns.

Keywords:

Stop–go traffic , Safe spacing , Behavioral change point , Behavioral patterns , Artificial neural networks , NGSIM data