Demand responsive transit scheduling method considering passenger travel choice

Demand responsive transit service faces the challenge of maintaining operating profits while complying with passenger preferences. To address this, a multinomial logit model was established to describe passenger travel choice behaviors, and a service-level-based pricing was developed. Subsequently, assortment optimization was integrated with transit scheduling to establish a dynamic scheduling model for demand responsive transit. An improved dynamic insertion algorithm was designed to efficiently solve the model. Finally, case studies were conducted using the Sioux Falls network for simulation analysis and the actual road network of Huangpu District in Guangzhou for practical analysis. The simulation case validates the feasibility of the model. Meanwhile, the practical case demonstrates that during peak periods, the model services 83.3% of passenger demands, and 90.0% during off-peak periods. The improved dynamic insertion algorithm can respond to passenger demands in seconds. Compared with the trip-based pricing, the service-level-based pricing increases operating profits and passenger service rates by 18.2% and 5.0%, respectively. Moreover, offering multiple service alternatives through assortment optimization increases operating profits and passenger service rates by 27.4% and 12.8%, respectively, while reducing average price by 6.3%. These results confirm that our proposed model outperforms the traditional scheduling model in terms of improving economic benefits and the number of passengers served.