Integrating bottom-up GIS and machine learning models for spatial-temporal analysis of electric mobility impact on power system

The ongoing electrification in the light-duty transportation sector represents a pivotal shift that deeply influences electricity distribution networks' operations, introducing a peculiar demand profile characterised by spatial and temporal variability. To address these challenges posed by the increasing adoption of Electric Vehicles (EVs), this work integrates a Machine Learning (ML) model for the vehicle selection procedure in a holistic Spatial-Temporal Model (STM) that accurately simulates the most typical stochastic behaviour within the transportation and electricity networks. The methodology assesses traffic behaviour, evaluates the grid impact of charging processes, and extends the analysis to flexibility services, particularly the provision of primary frequency regulation. The methodology is applied to the Lombardy region in Italy, adopting the 2030 e-mobility scenario defined by policymakers as a reference. This framework selects EVs diverting from linear probabilistic extraction models based on penetration rates by exploiting behavioural patterns and the socio-economic characterisation of EV drivers. Relying purely on open-source data, the work demonstrates the frequency regulation potential of EVs fostered by smart charging algorithms, which increase the power band available for grid services. The results of the procedure provide actionable insights for grid operators and urban planners, bridging the gap between transportation and electrical infrastructure.