摘要
候车时长是出租车乘客选择乘车点的重要判断依据,对实现人工智能趋势下的智慧交通具有重要意义。针对出租车乘客候车时间长、打车难等问题,提出一种利用作息时空特征优化神经网络的候车时间预测模型。该模型将出租车轨迹、城市兴趣点和时间作息片段等多源数据映射至50m*50m的精细地理网格中,以网格为单位,利用作息时空特征优化的神经网络对出租车行驶时空规律进行训练建模,从而预测在一定时空约束条件下空驶出租车的乘客候车时长。实验结果表明,精细的网格粒度和作息时空特征能帮助神经网络模型学习到更精确的运载规律知识,提高候车时长预测准确率。该方法为城市居民的智慧出行提供了科学合理的决策参考。
Taxi waiting time is important for taxi passengers to choose the pick-up points.The research on the taxi waiting time is of great significance for the realization of intelligent transportation driven by artificial intelligence technologies.Aiming at the problem of long waiting time in the passengers’taking taxis,this paper proposed a taxi waiting time prediction model based on work and rest spatio-temporal features-improved neural network.In this model,the multi-sourced data(e.g.taxi trajectories,city points of interest and time clips)is mapped into several 50m*50m-sized geogrids.Based on the grid fusion data,the spatio-temporal schemas of taxi driving can be learned by the neural network optimized by the work and rest spatio-temporal features.It can predict the taxi waiting time of the passengers under certain spatio-temporal environments.The experimental results show that the fine granularity grids and spatio-temporal schemas can effectively optimize the neural network model and improve the accuracy of taxi waiting time’s prediction.This method provides a scientific decision support for the intelligent travel of taxi passengers in the cities.
作者
雷永琪
李娜
陈智军
何渡
张雨昂
LEI Yong-qi;LI Na;CHEN Zhi-jun;HE Du;ZHANG Yu-ang(College of Computer Science and Information Engineering,Hubei University,Wuhan 430072,China;Hubei Acdemy of Science and Technical Information,Wuhan 430071,China)
出处
《软件导刊》
2021年第8期29-37,共9页
Software Guide
基金
湖北省教育厅科研计划项目重点项目(2019CFB757)
湖北省自然科学基金面上项目(2019CFB757)
湖北省教育厅科学研究计划重点项目(D20161001)
中央高校基本科研业务费专项资金资助项目(2019III050GX,2019III007GX)。
关键词
时空大数据
时空特征
神经网络
城市计算
多源数据
spatio-tempora big data
spatio-temporal features
neural network
urban computing
multi-source data
