A Hybrid and Adaptive Metaheuristic for the Rebalancing Problem in Public Bicycle Systems

To meet the fluctuating demand for bicycles and for vacant lockers at each station, employees need to actively shift bicycles between stations by a fleet of vehicles. This is the rebalancing problem in public bicycle systems. In this paper, we propose a new objective function to the rebalancing problem, which meets the actual circs better. Then we explore a new method combines data mining process with GRASP-PR which incorporate GRASP and path-relinking procedure to experiment, not a single activation, but multiple and adaptive executions of the data mining process during the metaheuristic execution. And some improvements are made in some phases of the algorithm according to the feature of the bicycle rebalancing problem. Practice examples and comparison with the typical algorithm in the fields are made. The results show that the new proposals were able to find better results in less computational time for the rebalancing bicycle problem. The research result has been implemented in Hangzhou, China.