停車場即時停車導引控制模式之建置

Abstract

摘要 為了改善停車導引的效率，過去已經有大量的研究與系統被用於導引駕駛至最近的停車位。然而，過去幾乎沒有相關的研究提出如何在一個大型的停車場裡導引駕駛去搜索可用的停車位。尤其是許多為了滿足過量的停車需求而建立的大型停車場。因此如何有效地使用方法在一個高度擁擠的停車場裡搜索到停車位是非常必要的。而憑藉著先進的技術(例如：e-Tag)，利用導引系統將駕駛導引至最近的停車空間，以減少尋位距離和步行距離是可行的。 有鑑於此。本研究旨在提出一個動態停車導引控制模式，它可以在系統設置成本與使用者行駛和步行成本之間作出一個最佳化的權衡。為此，本研究採用基因演算法找出最理想的停車位分群，並隨著分群的數量裝設偵測裝置，然後根據停車的需求，依序地開放停車位給駕駛，有效地限制駕駛的停車搜尋範圍。接著根據最佳分群結果，開發一套模糊控制器，動態地依據剩下的停車位以及車輛抵達狀況決定是否開放下一個分群。為了計算出駕駛的尋位距離與步行距離，本研究利用羅吉特模型模擬駕駛在開放的停車空間裡的停車選擇行為。 為了證明模型的適用性，本研究使用兩個案例，私人與公有停車場進行簡例分析。私人停車場的使用者通常為公司或工廠員工，目的是提供職員上班停放車輛，因此會有較長的停車時間(例如：8個小時)和相同的進入與離開時間。然而，公共停車場的使用者則是一般民眾，在停車時間上相對於私人停車場是較短的，而且使用者進入與離開停車場的時間也不會是一致的。結果證實，此模型可以找出停車場的最佳分群結果並提供有效的導引控制，減少駕駛的尋位與步行距離。另外，本研究可以透過設置成本和使用者成本的主觀權重設定，彈性地確定最佳的分群結果。

Abstract To facilitate the efficiency of parking searching guidance, numerous studies and systems have been conducted to guide drivers to the parking vacancy at the nearest parking lot. However, to guide drivers to search an available parking space in a large-scale parking lot, the related studies are barely proposed and applied. Especially, to meet the excessive parking demand, many large-scale parking lots have been built. How to search available parking spaces at a highly crowded parking lot at an efficient manner is essential. With advanced technologies (e.g. e-Tag), it is feasible that the system can efficiently guide users to nearest parking space to reduce their driving and walking distance. Based on this, this study aims to propose a dynamic parking guidance control model which can optimally trade-off the system installation cost and users driving and walking distance cost. To do so, this study employs genetic algorithms to optimally determine the zoning areas of total parking space along with the corresponding number of detection devices so as to sequentially open for parking users according to the parking demandand to effectively curtail the target searching area. Based on the optimal zoning, a fuzzy logic controller is developed to dynamically determine the open timing of each parking zone based the percentage of remaining vacant parking spaces and arrival drivers. A logit model is then used to simulate the driver behaviors in choosing among opened parking zones so as to compute the driving distance and walking distance. To demonstrate the applicability of the proposed model, two case studies of parking lots, a private and a public parking lot, are conducted. The users of the private parking lot is usually limited to employees who have long parking duration (e.g. 8 hours) and the same entrance and exit times. However, the public parking lot is open to the public those who have comparatively shorter parking duration and rather different entrance and exit times. The results confirm that the proposed model can optimally determine the zoning and provide useful guidance control to reduce drivers driving and walking distance. Additionally, the proposed model can flexibly determine the optimal zoning area according to the subjective weight settings of installation cost and user cost.