Estimation of Potential Charging Demand for Taxis and Spatiotemporal Feature Decomposition Based on Trajectory Data
In the transition of urban taxi fleets toward electrification in large cities,the charging demand of taxis exhibits characteristics of high charging load and strong spatiotemporal randomness,with a noticeable mismatch between charging supply and demand in time and space.To accurately estimate the potential charging demand post-the full electrification of the taxi fleet,this study introduces a bottom-up conceptual framework and a binary tree algorithm based on trajectory map matching,leveraging fuel taxi trajectory data exclusively.This approach provides a new paradigm for estimating charging demands in regions lagging behind in the electrification transition.The model and algorithm are validated using trajectory data from 890 electric taxis with battery status fields (State of Charge,SOC).Results show that the estimation errors of indicators such as the number of charging segments and charging amount are less than 6.5％.Moreover,the model also exhibits high spatiotemporal distribution estimation accuracy under different parameter settings (battery depletion threshold θ) and spatial scales (with grid sizes of 500 m,1000 m,and 10000 m),ensuring their applicability in real-world scenarios.Specifically,the temporal distribution error of charging amount is less than 8.5％ in the best-case scenario,and over half of the charging amount within 500 m grids has a spatial distribution error less than 0.3,with 59％ of the 500 m grids having an estimated error of charging segment count less than 0.3.Building upon this,the Singular Value Decomposition (SVD) algorithm is used to decompose and reduce the dimensionality of the spatiotemporal matrix of charging demands,identifying spatiotemporal patterns of potential charging demands within the real road network of Beijing's Sixth Ring Road at road level.Finally,a case study is conducted using trajectory data from 1913 taxis in Beijing over three consecutive days from March 9th (Monday) to March 11th (Wednesday) in 2019,and the results indicate that the spatial distribution of potential charging demands for taxis in Beijing exhibits prominent clustering features in key areas and critical corridors,corresponding to high-density charging demands associated with residents'high activity levels and long-distance travel.The decomposed charging demands reveal a spatiotemporal structural pattern dominated by regular charging demands,with supplementary heterogeneity in charging demands between morning and afternoon,as well as during working and non-working hours.This analysis method assists in uncovering the spatial distribution structural characteristics of potential charging demand and spatiotemporal coupling relationships,providing decision-making references for long-term planning of charging infrastructure,grid load scheduling,and charging demand management in the electrification transformation of taxi fleets.
urban traffictraffic electrificationcharging demandbinary treemap matchingsingular value decompositionspatiotemporal characteristicstrajectory data
NETL
NSTL
万方数据
