基于泊松分布的出租车合乘概率及等待时间建模

Method of Taxi Carpooling Probability and Wait Time Based on Poisson Distribution

为了反映城市出租车合乘的成功率,准确描述合乘的效果,利用出租车合乘概率和等待时间对合乘问题进行分析。依据泊松分布原理,讨论出租车合乘中出租车载客车辆的泊松分布,建立基于泊松分布的出租车合乘概率模型,并通过概率模型结合n重伯努利试验原理,构建出租车合乘等待时间模型,探讨出租车数量、空载率、平均行驶速率和行驶目的地分布对合乘概率和合乘等待时间的影响。以南京市雨花南路路段出租车行驶的GPS数据为研究对象,对模型进行验证。研究结果表明:出租车载客车辆数据通过显著水平为0.05的K-S检验,其分布服从泊松分布;出行的高峰时段,路段内出租车数量增大,空载率降低,对于行驶目的地分布相对集中的区域,出租车合乘的概率为60%~80%,合乘的等待时间为5~15min;出行的低峰时段,路段内出租车的数量减少,空载率增加,对于行驶目的地分布相对集中的区域,出租车合乘的概率为20%~60%,合乘等待时间在15min以上;在该路段内的出租车平均行驶速度相对稳定,对出租车合乘概率和等待时间影响较小;出租车合乘适宜在日常出行的高峰时段采用,可通过1次载客多次合乘的方式提高合乘效率;该模型具有一定的有效性和实用性,能够为交通管理部门和合乘者选择合乘提供参考。

To reflect the success rate of urban taxi carpooling,carpooling effects must be examined accurately and the problem of taxi carpooling must be analyzed in relation to carpooling probability and wait time.The Poisson distribution in taxi carpooling is discussed on the basis of the Poisson distribution principle.The probability model of taxi carpooling is established.A taxi wait time model is constructed on the basis of a probabilistic model and n-weight Bernoulli experimental principle.The number of taxis,no-load rate,average speed,and the distribution of travel destinations are found to influence carpooling probability and wait time.GPS data of taxis in Nanjing City are selected as a case study,and the model is verified.Experimental results showthat the vehicle data of taxi passengers obey the Poisson distribution,based on a K-S test with a significant level of 0.05.During the peak period of travel,the number of taxis increases,the noload rate decreases,the probability of taxi carpooling is 60%-80%,and the wait time for a carpool is 5-15 min for the relatively concentrated distribution of driving destinations.During the low peak period of travel,the number of taxis in the section decreases,the no-load rate increases,the probability of taxi carpooling becomes 20%-60%,and the wait time for a carpool becomes more than 15 min for the relatively concentrated distribution of the driving destination.The average speed of the taxi in this road section is relatively stable,which has little influence on the taxi carpooling probability and wait time.Taxi carpooling is suitable for daily travel during peak hours and can improve efficiency for repeated carpool routes. This result confirms the effectiveness and practicality of the proposed models and provides a reference point for traffic management departments and co-riders.