A Data Driven Urban Taxi Charging Station Locating And Sizing Method

As the environmental issues are gradually being valued,electric vehicles are favored as a low-carbon transportation tool.However,battery technology is still not fully mature,and the battery capacity is difficult to satisfy the daily needs of taxis.Meanwhile,the daily operation of taxis mostly uses a double-shift mechanism,taxis must be fully charged before handover to the next driver,and it is prone to queue up and wait for charging during the peak of handover.Therefore,reasonable charging station planning is essential for electric taxi groups.In fact,the operating behavior of electrical taxis is basically the same as that of traditional taxis.Based on this assumption,this article uses the historical driving trajectory data of taxis in Fuzhou City and excavates the behavior patterns of taxis.Based on the excavation result,charging station planning is implemented.Considering the preferential policies of low-cost charging in the off-peak hours and the characteristic of electric taxis tend to charge in duty shift hours,this paper proposed a duty shift recognition algorithm to mine the temporal and spatial distribution of duty shift event.According to the recognition results and the characteristics of Fuzhou taxi behavior pattern,combining with the driving conditions of Fuzhou,and comprehensively considering the interests of both the driver and the charging station,a planning plan for the future electric taxi centralized charging station in Fuzhou was proposed.The final planning integrates multi-dimensional information such as driver behavior patterns,duty shift behavior,road service capability,and social benefits,which is verified to be reasonable.The main contributions are illustrated as follows:(1)This paper proposed a taxi’s duty shift recognition method,it combines the DBSCAN algorithm and kernel density analysis method to identify the spatial and temporal information of the duty shift event.The experiment has been verified a scientific and effective method through sampling investigation.(2)A topological model of the Fuzhou road network is firstly established,then,the taxi’s traffic flow and driving speed of each time period and road section in Fuzhou city is calculated using the prior knowledge from the historical trajectory data.The OD reverse technology is applied to generate 24-hour OD matrices according to the taxi’s traffic flow and road network.The experiment shows the result is consistent with traffic pattern in Fuzhou.(3)Four driving cycles representing various traffic condition are developed using the Markov model and driving speeds in each road section.A vehicle kinetic model is employed to estimate the power consumptions of four cycles,the estimation results can reflect the real world traffic conditions in Fuzhou.(4)Based on Monte Carlo sampling method and multi-dimension information obtained by previous analysis,this study treated the spatial and temporal distribution of duty shift events as the initial distributions to simulate the electrical vehicles behavior in 24 hours,and further predicted the spatial and temporal demand of charging behaviors.The simulation is implemented under the real world parameters of electrical vehicles.(5)According to the spatial and temporal distributions of charging demand,this study comprehensively consider the interests of drivers and charging stations,an overall cost model is established.Then,the simulated annealing algorithm is applied to find out the optimal planning for the charging stations.The result shows the proposed method can automatically optimize the planning of charging stations,and the service sections of each station are clearly divided,which is anticipated to effectively alleviate the pressure of charging.