考量使用者偏好之高速公路服務區充電樁設置最佳化

Abstract

為因應全球暖化所帶來之極端氣候等影響，電動車已成為最主要之推廣運具 以減少二氧化碳於交通領域之排放量。而面對近年來電動車逐年增長之市占率， 勢必需要提供完善之充電服務以滿足未來不斷上升之需求。針對電動車較短之行 駛里程以及長時間之充電過程等相關特性，長途行駛的充電需求及使用者在充電 期間之時間價值差異性需進行充分討論，進而產生高速公路路網中充電站之設置 問題，並考慮電動車使用者對不同充電站間之偏好。 為解決該問題，本研究提出了一考量使用者均衡之雙層模型，以描述使用者 路徑選擇行為以及產生充電站位置及容量的最佳化設置決策。隨後，將該模型合 併為單層模型，以加強其尋找最佳解的能力。此外，本研究建立局部搜索和元啟 發式等演算法作為處理大規模問題之替代方案。這些方法被應用於一個簡易案例 中，以比較其解決問題之效率及效用。結果顯示，單層模型表現最佳，而局部搜 索和 ABC 算法亦能夠在較短的計算時間內產生良好之解決方案。另外，本研究也 針對不同假設情境進行敏感性分析，顯示電動車使用者偏好對充電站設置之結果 產生顯著之影響。最後，本研究套用一實際案例，以驗證該模型的應用於真實情 境之可行性。 面對未來充電需求之成長之議題，本研究提供一最佳化模型，協助決策者尋 找高速公路路網中充電樁設置之最佳方案，並考量電動車使用者之充電偏好，使 模型能充分模擬駕駛在選擇充電站之決策行為，提升電動車使用者於充電期間之 滿意度。該成果有助於持續推動電動車之普及，進而實現未來淨零排放之目標。

Electric vehicles (EVs) have emerged as a key solution for reducing carbon dioxide emissions in the transportation sector, addressing the exacerbation of global warming. With an observable growth of the EV adoption, comprehensive charging services must be provided to to meet the anticipated surge in demand. Given their shorter driving range and longer charging times compared to conventional vehicles, it is essential to focus on the charging demand of EV users traveling long distances and their value of time during charging. This necessitates addressing the problem of charging station deployment in highway networks, considering EV users' preferences among stations.

To solve this problem, this study proposed a bi-level model involving user equilibrium to describe route choice behavior and the optimization of deployment decisions for location and capacity. The model was then combined into a single level to enhance its ability to find optimal solutions. Additionally, algorithms such as local search and meta-heuristics were established as alternatives for solving large-scale problems. These methods were applied to a toy network to compare their effectiveness and efficiency. The results showed that the single-level model performed the best, while the local search and Artificial Bee Colony (ABC) algorithms could generate great solutions with shorter computational times. A sensitivity analysis was also conducted to demonstrate the significant impact of EV users' preferences on the final solution for charger distribution. Finally, a real-world case was established to verify the practical application of the model.

With the predictably increasing charging demand in the future, this study provides a useful tool for decision-makers to determine charger installation in each service area along highway networks. A specific focus on the preferences of EV users enables the model to capture practical driver behavior in the selection of charging stations, optimizing satisfaction with charging duration. This support for the continued promotion of EV adoption contributes to achieving the net-zero emission goal in the future.