物聯網架構下運用路程與停車場車流量實現停車推薦系統

Abstract

在現今資訊爆炸且硬體發展日益先進的時代中，使用物聯網技術以其硬體與軟體架構進行應用並達到智慧城市目的之相關開發與建置正快速地發展中，而其中都市利用多種不同感測器與資料來進行停車行為的優化則成為相當熱門的議題，尤其在人口密集區域大量停車需求與眾多停車區域的環境中，停車行為的分配顯得更為重要。\n在本篇論文當中，我們建立出一套較符合真實情況的停車推薦系統解決都市中有停車需求者無法有效的利用城市中剩餘的停車空位，同時也可以避免都市中原已壅塞的交通量更加惡化，降低城市車流量負載，本研究透過將停車行為拆分成兩個部分並運用等候理論的公式模型推導進行真實情況的預測，第一部分為估計出有停車需求的使用者發出停車請求後直到經過系統分配到達推薦的停車場中間路途所需耗費的時間，並且將其時間用於第二部分預估停車場在經過路途預估所需時間後能保有空位之機率，藉此達到良好的停車成功率，減少車輛在經過漫漫長路到達後卻無位可停的機率\n最後使用SUMO交通模擬器進行實驗模擬，結果顯示在壅塞的都市交通環境中，使用本論文方法相較於傳統的決策方法能有效降低使用者停車總花費時間與到達後因車位已滿而失敗的機率。

In the age of information explosion and advanced hardware development, the related exploitation and construction of using IoT (Internet of Things) technology with its hardware and software framework application to achieve the purpose of “smart city” is developing rapidly. The optimization of parking behavior using a variety of different sensors and data in cities has become a very popular and important issue. Especially in the environment of a large number of parking needs in densely populated areas with many parking areas, the allocation of parking behaviors is more important.\nIn this thesis, we propose a parking recommendation system that conforms with the real situation to solve the parking problem by effectively using remaining parking space in the city. At the same time, it can also avoid worsening the traffic jam and thus, balance the traffic flow in the city. This research proposes the recommendation method by analyzing the parking behavior with two parts, and using the model of queuing theory. The first part is to estimate the time consumption from the parking request generated by the user to reaching the candidate parking lot. The time consumption from the first part will be used in the second part to calculate the probability of getting a free parking then. By this calculation, we can expect to achieve the better parking success rate,and avoid the cars to circle around for free parking space.\nTo validation our model, we use the SUMO traffic simulator for experimental simulation. The results show that in a congested urban traffic environment, our method can effectively reduce the total time spent on parking and the failure probability of the parking space after arrival.