隨著電動車於全球交通運輸體系中快速普及，充電需求之成長已逐漸成為影響使用便利性與整體運輸效率的重要因素。相較於傳統燃油車輛依賴成熟之加油站網絡，電動車充電具有補能時間較長、需求受通勤與活動型態影響明顯，以及時間與空間分布高度不均等特性。現行充電服務多以固定式充電站為主，然而其建置與營運受土地取得、電力容量與投資成本等限制，難以即時因應需求波動，導致尖峰壅塞與離峰閒置並存之現象，降低整體系統效率。 本研究聚焦於充電需求具高度時空不均特性之情境，探討電動車充電服務之資源配置與動態調度問題。首先分析充電需求特性與既有充電模式之限制，進而建立充電服務配置與路徑決策之數學模型，將其建構為結合動態需求與資源限制之序列決策問題。為提升決策彈性與因應需求不確定性，本研究引入強化學習與最佳化方法，建構具調度能力之決策架構，以提升資源利用效率並降低服務延遲。 透過模擬實驗比較不同策略於多種需求情境下之績效表現，結果顯示，具動態調度能力之方法相較於傳統靜態配置策略，能有效改善資源分配效率與整體服務品質。本研究成果可為電動車充電服務之規劃與營運管理提供參考，並有助於提升充電系統效率與永續運輸發展。;With the rapid growth of electric vehicle (EV) adoption， charging demand has become a critical factor affecting user convenience and overall transportation efficiency. Unlike conventional fuel vehicles supported by a mature refueling network， EV charging involves longer service times and significant spatiotemporal demand imbalance. Existing charging systems primarily rely on fixed charging stations， whose deployment is constrained by land availability， grid capacity， and investment costs. These limitations often lead to congestion during peak periods and underutilization during off-peak hours， reducing overall system efficiency. This study investigates the resource allocation and dynamic scheduling problem of EV charging services under uneven demand conditions. A mathematical model is developed to formulate charging service configuration and route decisions as a sequential decision-making problem with dynamic demand and resource constraints. Reinforcement learning and optimization approaches are incorporated to enhance scheduling flexibility， improve resource utilization， and reduce service delays. Simulation results demonstrate that dynamic scheduling strategies outperform traditional static allocation methods in improving service efficiency and overall system performance. The findings provide practical insights for EV charging service planning and contribute to the development of more efficient and sustainable transportation systems.
