物聯網驅動之邊緣運算 V2G 排程：兼顧電網穩定性與經濟效益之研究

Abstract

隨著6G網路的到來與物聯網驅動的智慧社區快速興起，電動車正逐漸成為未來能源系統的核心組成。車輛對電網（V2G）技術使電動車能夠作為可動態與電網交換電力的移動儲能單元，從而提升整體能源靈活性與穩定性。然而，實務應用中仍面臨多項挑戰：缺乏協調的充放電會導致電網負載劇烈波動；電網資源容量有限；高熱應力下電池退化加速；以及集中式排程架構的通訊與運算延遲問題。為解決上述困境，本文提出一種結合物聯網與邊緣運算的兩階段電動車充放電排程框架。該框架首先在「成本最小化階段」中，聯合優化動態電價與溫度調整後之電池退化成本；接著在「基於代替最優解的平滑階段」中，於維持近似最優成本的前提下細化基線解以增強排程連續性；外層迭代機制則依據電池溫度模擬結果動態更新退化成本權重，確保決策與熱狀態一致。整個框架部署於階層化邊緣運算架構中，使社區控制節點能在低延遲且保護用戶隱私的條件下執行本地化優化。於真實家庭負載與電動車需求場景下的模擬結果顯示，相較於單階段平滑約束排程方法與僅進行成本優化的基線方法，所提兩階段方法在降低經濟成本、維護電池健康與提升電網穩定性方面展現更優的折衷效果。本研究成果突顯了結合物聯網、邊緣運算與先進優化技術，打造可擴展且具實務可行性的6G智慧社區V2G解決方案之潛力。

The advent of 6G networks and the proliferation of IoT-enabled smart communities are accelerating the deployment of electric vehicles (EVs) as integral components of future energy systems. Vehicle-to-Grid (V2G) technology enables EVs to act as mobile storage units that can dynamically exchange power with the grid, enhancing energy flexibility and stability. However, practical challenges remain, including grid load fluctuations due to uncoordinated charging, limited grid resource capacity, battery degradation under high thermal stress, and the latency limitations of centralized scheduling frameworks. To address these issues, this thesis proposes a novel two-phase EV charging and discharging scheduling framework integrated with IoT and edge computing architectures. The proposed method first formulates a cost minimization phase that jointly optimizes dynamic electricity cost and temperature-adjusted battery degradation cost. It then applies an alternative optima-based smoothing phase to refine the baseline solution for operational continuity while maintaining near-optimal cost performance. An outer iteration mechanism further updates degradation cost multipliers based on battery temperature simulation, ensuring thermal-consistent decision making. The framework is deployed within a hierarchical edge computing architecture, enabling community control nodes to perform localized optimization with low latency while preserving user data privacy. Simulation results under realistic household load and EV demand scenarios demonstrate that the proposed two-phase approach achieves superior trade-offs among economic cost reduction, battery health preservation, and grid stability enhancement compared to both single-phase smoothing-constrained scheduling and baseline cost-only optimization. These findings highlight the potential of combining IoT, edge computing, and advanced optimization to build scalable and practical V2G solutions for future 6G-enabled smart communities.