Analysis of charging demands, influencing factors and spatial effects of electric vehicles based on multi-source data and local spatial models

In order to alleviate the difficulty of charging caused by the popularization of electric vehicles (EVs), this paper conducts an in-depth analysis of the charging demands and influencing factors. According to the spatial distribution of charging demands, the target area is divided into more Voronoi unequal polygons, and the spatial characteristics of the built environment and socio-economic factors is qualitatively analyzed. Then, multicollinearity and spatial correlation tests are employed to eliminate redundant factors and explore the spatial clustering and correlation of charging demands and impact elements. A multi-scale geographically weighted regression and spatial autoregressive (MGWR-SAR) model are proposed to investigate such complex properties. An empirical study in Chongqing, China has shown that the charging demands in adjacent units exhibit obvious high-high and low-low clustering patterns, and are significantly influenced by the EVs with low SOCs, population density, parking lots density, transportation conditions, etc. The spatial impact degrees and scales vary with the factors and intervals, wherein the spatial scales of population density and road network density are local, with strong spatial heterogeneity; EVs with low SOCs, land use mixing and housing prices are close to global impacts. The spatial dependence of charging demand in high demand areas and charging peak periods is stronger than that in low demand and off-peak. There are spatial dependence and heterogeneity in charging demands and influencing factors, which makes MGWR-SAR superior to other models. These findings will provide support for predicting charging demands and optimizing charging stations.