The emergence of multi-source and mass travel data and the gradual maturity of acquisition methods make the realization methods of travel demand more diversified. The on-demand bus is a pattern that can meet the needs of personalized and high-quality travel. This project proposes an on-demand bus system based on response to random users' real-time requests. Under the condition of fixed origin points and destination points, taking the minimum total waiting time of passengers and the maximum profit of the bus as the goals, a model based on the variables of bus fare discount and request valid time is established, and LINGO and the actual car-hailing data are used to calculate and check the model. Through calculation results and the comparison with the traditional bus and the express, the on-demand bus system has shorter travel time than the traditional bus and costs less than the express. Thus the research can be thought to provide a scientific basis for the establishment of the on-bus bus system, and provide theoretical support for further research
Papers Presented at the 2018 37th Southern African Transport Conference 9-12 July 2018 Pretoria, South Africa. Theme "Towards a desired transport future: safe, sufficient and affordable".