Self-balanced bicycle rental systems under dynamic demand forecasting

In this paper we develop a dynamic pricing framework for bicycle rental system to help achieve self-balance and apply this model to a simple two stations case. The first part is short-time demand forecasting for Origin-Destination (OD) pairs, using an Adaptive Difference Method. Such method integrates the information from both historical and real-time data to improve the precision of demand forecasting and estimation. Based on given OD demand, the second part illustrates a dynamic pricing model for self-balanced bicycle rental systems. We develop an optimum programming whose object is to achieve the self-balance of rental system under minimized society cost. We also employ discrete choice models to model the equilibrium between different travel modes. Lastly, this OD demand forecasting method and the dynamic pricing model are simulated in a two stations case.