基于使用行为分析的共享单车管理优化研究  

Management and Optimization of Shared Renting Bicycle Based on User Behavior Analysis

作　　者：傅哲 辛泓润 余力[1] 徐冠宇 FU Zhe;XIN Hongrun;YU Li;XU Guanyu(School of Information,Renmin University of China,Beijing 100872,China;International School,Bejjing University of Posts and Telecommunications,Bejjing 100876,China;Xu Teli School,Beijig Institute of Technology,Beiing 100081,China)

机构地区：[1]中国人民大学信息学院,北京100872 [2]北京邮电大学国际学院,北京100876 [3]北京理工大学徐特立学院,北京100081

出　　处：《信息系统学报》2019年第2期81-94,共14页China Journal of Information Systems

基　　金：国家自然科学基金项目(71271209,71331007)。

摘　　要：近年来,共享单车作为一种典型的共享经济应用的代表,越来越受到大众的欢迎,如何基于用户的使用行为分析来对共享单车管理进行优化是未来共享单车发展的重要问题。本文以2016-2017年纽约市的Citi Bike共享单车的使用作为研究数据集,从时间和空间的维度洋细分析全市各区域内共享单车的使用行为及特点;采用长短期记忆(long short-term memory,LSTM)神经网络预测分析各站点在高峰时段的共享单车存1和净流入量情况;在此基础上,针对共享单车时空的分布失衡问题,采用运筹学中的运输问题模型,研究高峰时段的调度策略,探索共享单车在各站点间调度的最小代价方案,以缓解共享单车时空分布失衡的问题。本文研究对优化共享单车管理具有一定借鉴意义。In recent years,as a ty pical representative of sharing economy application,renting bicycle is more and more popular.It is key isue about how to analyze the behavior of users based on the use of shared bicycle to improve their management.In this paper,we will focus on the problem of unbalanced distribution of bicycles in the region.We will use LSTM network to train and forecast the data of Citi Bike and use the minimum element method to find the most minimum cost scheme.The rescarch contents includes the follows:Exploring the distribution characteristics of shared bicycles in time and space;Using LSTM neural network to predict the stock and net inflow of shared bikes in stations during peak hours;Based on the forecast of the use of bicycles in each station,finding the scheduling strategy of peak season.Through the study of the results of the New York Citi Bike,we can not only further optimize the operation of the bike sharing management and alleviate the problem of unbalanced distribution of sharing bicycle,but share the experience to operation of China's bike sharing system as well.

关 键 词：共享单车 管理优化 LSTM 神经网络 运输模型 预测 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]