Solving a static bike repositioning problem using the artificial bee colony algorithm

Abstract

This study tackled a static Bike Repositioning Problem (BRP) with broken bikes. It aims to design the route and corresponding loading/unloading quantities for the reposition vehicle to satisfy the demands of all the stations in a bike sharing system. The objective is to minimize the CO2 emissions of the vehicle during repositioning. An artificial bee colony (ABC) algorithm with adaptive neighborhood operators is developed to search for the vehicle route, incorporating a newly introduced evaluation method. This evaluation method is designed for adjusting the loading/unloading quantities for the routing problem that allows for multiple visits and has a vehicle load related objective function. The performance of the proposed algorithm on the BRP is evaluated in different instances having 10-300 stations. The results show that the proposed algorithm can find the optimal solutions for small-sized networks within a much shorter time than the exact method. In addition, for the instances that the exact method cannot provide the optimal solution within the time limit, the proposed method outperforms both the genetic algorithm and the memetic algorithm. The result confirms the ability of the ABC algorithm in solving this type of problems.