Scholarlink

<p>Rapid electric vehicle (EV) adoption in car-dependent, urbanizing regions demands rigorous spatial planning for charging infrastructure, yet existing site-selection frameworks lack triangulated validation and rarely integrate accessibility analysis with suitability modelling. This study develops a GIS-based Multi-Criteria Decision-Making (MCDM) framework for EV charging station (EVCS) siting in Greater Doha, Qatar, integrating 20 spatial, infrastructural, and socio-economic variables across five domains against 45 existing EVCS locations, with suitability maps generated using Equal Weight, Analytical Hierarchy Process (AHP), Fuzzy AHP (FAHP), and TOPSIS, independently validated by three ensemble machine learning classifiers (Random Forest, Gradient Boosting Machine, and XGBoost). All three ML models confirm strong convergence with FAHP expert weights (RF: <i>ρ</i> = 0.802; GBM: <i>ρ</i> = 0.968; XGBoost: ρ = 0.974; all <i>p</i> &lt; 0.001) and very high inter-model agreement (<i>ρ</i> &gt; 0.96). Petrol stations emerge as the single most influential siting criterion across all ML models (21.9–26.3%), with recreational and infrastructure factors dominating overall suitability. FAHP outperforms all other methods, achieving a Kappa of 0.646 and ROC–AUC of 0.756, correctly identifying 82.2% of existing stations in high-suitability zones; AHP follows closely (Kappa = 0.618, AUC = 0.741, 80.0% accuracy), with both methods consistently outperforming TOPSIS. Threshold sensitivity analysis confirms that these rankings are stable across varying suitability-class definitions. Service-area modelling reveals critical accessibility gaps in Al Daayen and Al Wakrah, where substantial residential populations fall outside 10–15 km catchments. These findings recommend a hybrid deployment strategy combining dense EVCS networks around high-demand hubs in central Doha with targeted expansion into underserved municipalities, directly supporting Qatar's target of 1,200 chargers by 2030 and offering a replicable planning template for arid, rapidly urbanizing regions globally.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Spatial suitability and accessibility planning for EV charging infrastructure in Qatar: a GIS-MCDM approach validated by ensemble machine learning

  • 期刊论文

  • 发表日期：

    2026年4月15日

查看全文
  • Ammar Abulibdeh

摘要

Rapid electric vehicle (EV) adoption in car-dependent, urbanizing regions demands rigorous spatial planning for charging infrastructure, yet existing site-selection frameworks lack triangulated validation and rarely integrate accessibility analysis with suitability modelling. This study develops a GIS-based Multi-Criteria Decision-Making (MCDM) framework for EV charging station (EVCS) siting in Greater Doha, Qatar, integrating 20 spatial, infrastructural, and socio-economic variables across five domains against 45 existing EVCS locations, with suitability maps generated using Equal Weight, Analytical Hierarchy Process (AHP), Fuzzy AHP (FAHP), and TOPSIS, independently validated by three ensemble machine learning classifiers (Random Forest, Gradient Boosting Machine, and XGBoost). All three ML models confirm strong convergence with FAHP expert weights (RF: ρ = 0.802; GBM: ρ = 0.968; XGBoost: ρ = 0.974; all p < 0.001) and very high inter-model agreement (ρ > 0.96). Petrol stations emerge as the single most influential siting criterion across all ML models (21.9–26.3%), with recreational and infrastructure factors dominating overall suitability. FAHP outperforms all other methods, achieving a Kappa of 0.646 and ROC–AUC of 0.756, correctly identifying 82.2% of existing stations in high-suitability zones; AHP follows closely (Kappa = 0.618, AUC = 0.741, 80.0% accuracy), with both methods consistently outperforming TOPSIS. Threshold sensitivity analysis confirms that these rankings are stable across varying suitability-class definitions. Service-area modelling reveals critical accessibility gaps in Al Daayen and Al Wakrah, where substantial residential populations fall outside 10–15 km catchments. These findings recommend a hybrid deployment strategy combining dense EVCS networks around high-demand hubs in central Doha with targeted expansion into underserved municipalities, directly supporting Qatar's target of 1,200 chargers by 2030 and offering a replicable planning template for arid, rapidly urbanizing regions globally.