Skip to content
Open access

City-Scale Optimization of Public Electric Vehicle Charging Infrastructure: Spatio-Temporal Demand Forecasting, Graph Learning and Multi-Objective Siting

Sep 2026 · World Electric Vehicle Journal · 0 citations · 45 references

TL;DR

A multi-domain machine learning framework that integrates spatio-temporal charging demand, station-level infrastructure characteristics, land-use information, and graph-based network relationships for EV charging prediction and infrastructure siting is proposed.

Abstract

Electric vehicle (EV) adoption is accelerating rapidly, increasing pressure on public charging infrastructure, urban energy systems, and network-expansion planning. However, many existing studies examine charging demand forecasting, station network characteristics, or infrastructure siting as separate problems, limiting their ability to translate observed charging behaviour into coordinated planning decisions. To address this gap, this study proposes a multi-domain machine learning framework that integrates spatio-temporal charging demand, station-level infrastructure characteristics, land-use information, and graph-based network relationships for EV charging prediction and infrastructure siting. The analysis uses 8,544,695 public charging transactions recorded at 8553 stations across Beijing during January and July 2025. Charging sessions are aggregated into an hourly station panel and modelled using seasonal-naïve and historical mean baselines, ridge regression, gradient-boosted trees, graph-augmented gradient boosting, and a gated recurrent unit network. Model reliability is assessed through temporal validation, feature ablation, spatial holdout testing, cross-season transfer analysis, explainability, and non-parametric statistical comparison. The results show that graph-augmented gradient boosting achieved the strongest RMSE and variance-explained performance, with an RMSE of 53.80 kWh and R2=0.734, while the gated recurrent unit produced the lowest MAE of 23.82 kWh. Graph neighbour features yielded only a marginal and statistically non-significant forecasting improvement, indicating that the station network is structurally informative but predictively redundant once temporal history is available. For infrastructure expansion, NSGA-II achieved the highest Pareto front hypervolume and identified a knee-point solution of 156 additional chargers, reducing unmet demand by 43.4%. These findings demonstrate that integrated forecasting, graph analysis, and multi-objective siting can provide more defensible and operationally relevant evidence for city-scale EV charging planning.

Read PDF

Similar papers

#reinforcement learning Open access Sep 2026

Equity-Aware Planning of Urban Ultra-Fast Electric Vehicle Charging Infrastructure with Spatial Graph Learning and Grid Constraints

Urban ultra-fast-charging planning must connect equitable user service with physically deliverable charging power. This study formulates a heterogeneous graph linking demand zones, roads, candidate stations, and distribution buses, together with an infrastructure model for station locations, port-level power tiers, int...

Han Wu, Lu Shi, Xiao-Hang Gu et al. · 0 citations
Open access Sep 2026

Phased Planning of Expressway EV Charging Infrastructure: A Data-Driven Systems Approach Considering Spatial Heterogeneity and Peak Demand

Planning charging facilities at expressway service areas requires joint consideration of traffic growth, charging behavior, temporal demand peaks, charger technology, existing equipment, and local power-supply capacity. However, existing planning approaches rarely integrate spatially heterogeneous charging behavior, pe...

Yue-Li Guo, Xu-Ze Zhang, Hua-Yang Yu · 0 citations
#reinforcement learning Open access Sep 2026

Urban electric vehicle fast charging load forecasting: an LGRL approach

Test results show that the proposed hybrid forecasting method outperforms traditional algorithms in forecasting accuracy, convergence speed, and robustness under extreme conditions such as data missing, and it can effectively quantify the impact of load fluctuations on the voltage of distribution networks.

Hao-Min Jiang, Dan-Xiong Fei, Qiang Xing et al. · 0 citations
Conference Open access 2026

Electric Vehicle Charging Demand Prediction in Shenzhen Based on Random Forest and LSTM

Reliably forecasting how much energy electric vehicles (EVs) will require at public chargers is a prerequisite for infrastructure planning and matching available capacity to real-world usage. This study compares two representative approaches, Random Forest and Long Short-Term Memory (LSTM), to forecast hourly charging...

Hao-Ran Luo · 0 citations
Open access Sep 2026

Spatiotemporal digital twin for city-scale EV charging infrastructure using LSTM-GNN fusion and resilience-driven optimization

A holistic AI-driven digital twin framework that enhances grid stability, supports sustainable EV integration, and enables scalable deployment of future smart charging infrastructure is presented.

Deepa Somasundaram, B. Ravisankar, Chavvakula Janaki Devi et al. · 0 citations
Nov 2026

Spatial Heterogeneity and Scale-Dependent Determinants of Public EV Charging Station Utilization: An Integrated Semantic-Spatial Approach

The growing gap between the rapid expansion of public electric vehicle (EV) charging infrastructure and its realized utilization efficiency poses a critical challenge to sustainable urban mobility. Existing studies’ reliance on macrolevel aggregates or single-dimensional metrics fails to capture fine-grained spatia...

Peng-Fei Zhao, Xu-Yang Li, Jian-Xin Lin et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.