Unsupervised learning for deploying smart charging public infrastructure for electric vehicles in sprawling cities.

This paper presents a novel methodology to study the deployment of public smart charging stations (CS) of electric vehicles (EV) in a sprawling Latin American city. A relevant difference between developed and emerging economies is the reduced access to home charging in emerging economies, which is the case in Latin American cities. Thus, developing public charging stations represents a crucial factor in the mass adoption of EVs by road commuters. We develop herein a methodology for optimizing the deployment of smart charging stations under the sprawling phenomenon perspective. Our method comprises two steps. In step one, we applied principal component analysis (PCA) to facilitate the analysis of a sprawling city, and then we define candidates for potential locations from 'demand clusters' within an urbanized area, by K-means clustering analysis. In the second step, a stochastic programming model was employed to optimize the integration of infrastructural facilities with distributed energy resources (DERs) and EV charging stations using a collaborative strategy to minimize its energy consumption cost under demand uncertainty. We demonstrate the capabilities of this approach through a case study in the city of Lima. Experimental results reveal managerial insights for different stakeholders (i.e., government, industry, academia, and civil society) to promote policies, investment, and incentives. • Explores the EVs public charging stations deployment under the urban sprawl phenomenon perspective. • Principal component analysis (PCA) guided K-means clustering defines energy clusters. • Demonstrates the superiority of the PCA-based electromobility integration strategy. • Managerial insights are made to deploy EVs smart charging stations under uncertainty. [ABSTRACT FROM AUTHOR]