Cellular Automata Model for Analysis of Lane-Changing Behavior on Damaged Pavement

Abstract

Drivers often perform lane-changing behavior for driving benefits on a broken road. Accordingly, a lane-changing model for two-lane traffic under pavement damage conditions was established based on the cellular automata NaSch model by introducing slow start and new lane-changing rules. To simulate the characteristics of a driver, traffic flow, and lane change under different road conditions, a simulation process was established in four steps, namely, lane-changing demand, lane selection, clearance detection, and lane change execution. Road damage level is classified from the perspective of vehicle operation. Vehicle driving benefit is calculated through utility theory to establish a lane selection model. A lane-changing coefficient is also introduced to study the effect of pavement damage on vehicle operating characteristics. The driver is divided into four categories based on the differences among drivers' behavior, namely, adventurous, alert, cautious, and slow. The characteristics of the different types of drivers under pavement damage conditions were analyzed by setting simulation parameters. Results showed that the lane-changing coefficient would increase with the increase of pavement damage level. A high level of damage would result in a low benefit of the vehicle traveling on damaged roads, which further increases the probability of lane change that can properly simulate the effect of pavement damage on vehicle lane-changing behavior. The highest lane-changing rate was obtained by the adventurous driver in medium density, and the speed variance and lane-changing rate gradually increases with the increase of damage level. The high lane-changing rate also indicates that pavement distress aggravates lane-changing behavior and interferes with the normal flow of traffic operation, thereby providing a theoretical basis for strengthening traffic safety management of damaged road sections.