基于多层次网格划分算法的出租车全局调度策略的研究与实现  

Research and implementation of taxi global scheduling strategy based on multi-level meshing algorithm

作　　者：徐方方 戴大蒙[1] Xu Fangfang;Dai Dameng(School of Computers&Artificial Intelligence,Wenzhou University,Wenzhou Zhejiang 325035,China)

机构地区：[1]温州大学计算机与人工智能学院,浙江温州325035

出　　处：《计算机应用研究》2021年第1期115-119,共5页Application Research of Computers

基　　金：国家自然科学基金青年科学基金资助项目(KZ1511040);国家科技支撑计划资助项目(KZ1611036);2019年浙江省大学生科技创新活动计划暨新苗人才计划温州大学项目(KZS1910041);温州市551人才计划资助项目(R20180520)。

摘　　要：城市单元格划分算法应用于出租车调度方法时无法解决山川、河流、大型建筑等天然屏障对距离计算的负面影响,更无法确定网格之间的稳定性。针对此问题,提出了一种面向复杂地理环境的多层次网格划分算法,采用历史数据拟合出两点之间行驶距离的精确值,替代了传统路径计算方法,为距离计算问题提供了新的解决思路,多层次划分更使网格的稳定性得以保证。再结合线性规划方法,辅以时序图和散点图等时空特征识别技术,以高效益和负载均衡为目标,对空载出租车进行实时调度。实验结果表明,该方法提升了整体效益,缩短了乘客打车等待时间,达到了经济效益和社会效益的双提升,具有普适性和广泛的应用前景。The urban cell division algorithm applies to the taxi scheduling strategy cannot take into account the negative impact of natural barriers such as mountains,rivers and large buildings on distance calculation,nor can it guarantee the stability between grids.As to these problems,this paper proposed a multi-level gridding algorithm for complex geographic environment,and used the historical data to obtain the exact distance value between two grids by fitting,which replaced the traditional path calculation.The method also provided an innovative solution to the problem of distance calculation between grids.It was combined with the linear programming method,supplemented by time-space and featured recognition techniques such as time series and scatter plots,with high-profits and traffic load balancing targets,real-time scheduling of empty taxis.The experimental results show that the proposed method improves the overall benefits,shortens the waiting time for passengers,thus achieving a double improvement in economic and social benefits.And it is universal and has broad application prospects.

关 键 词：多层次网格划分 全局调度 GPS大数据 时空特征 供需均衡 

分 类 号：U491[交通运输工程—交通运输规划与管理]