基于网络模型的城市公共自行车需求量预测研究

发布时间：2018-05-08 18:20

本文选题：自行车共享系统 + 分层聚类算法　；参考：《计算机应用研究》2017年09期

【摘要】：自行车共享系统逐渐出现在许多城市中,由于在不同时间和站点的自行车需求量(租/还量)不平衡,系统中各站点的自行车需要人工频繁地调整使其不断达到平衡状态,然而实时监控并不能很好地解决这个问题。因此,提出了一个基于网络图的预测模型,可以预测未来时间段内的某个站点自行车的需求量,提前对站点自行车进行分配。通过分层聚类算法对预测站点进行聚类,得到与其相关的站点簇,并对站点簇构建网络模型。最后,使用纽约(NYC)和华盛顿(D.C.)两个自行车共享系统的数据进行实验,并与基线法、历史平均法及ARIMA模型进行比较。结果发现同一簇的站点具有相似的使用模式,模型预测误差率不高于0.45,网络模型预测性能较好,且能够应用于不同城市的自行车共享系统。
[Abstract]:Bicycle sharing system is gradually appearing in many cities. Because of the imbalance of bicycle demand (rent / amount) at different time and site, the bicycle needs to be constantly adjusted by manual and frequent adjustment in the system. However, real-time monitoring can not solve this problem well. Therefore, a network based on network is proposed. The forecast model of collaterals can predict the demand of bicycle in a certain site in the future time, allocate the bicycle in advance, cluster the forecast site by hierarchical clustering algorithm, get its related site cluster, and build the network model of the site cluster. At the last, use two self-propelled New York (NYC) and Washington (D.C.). The data of car sharing system are tested and compared with baseline method, historical average method and ARIMA model. The results show that the same cluster of sites have similar usage patterns, the prediction error rate of the model is not higher than 0.45, the network model has better prediction performance and can be applied to the bicycle sharing system in different cities.

【作者单位】：南京师范大学计算机科学与技术学院;江苏省信息安全保密技术工程研究中心;
【基金】：国家自然科学基金资助项目(41171298)
【分类号】：TP311.13

【相似文献】