完整後設資料紀錄
DC 欄位語言
dc.contributor.author蔡怡玟zh_TW
dc.contributor.author楊茗婷zh_TW
dc.contributor.author陳玫婷zh_TW
dc.date109學年度第二學期zh_TW
dc.date.accessioned2021-10-14T06:45:22Z-
dc.date.available2021-10-14T06:45:22Z-
dc.date.submitted2021-10-14-
dc.identifier.otherD0716278、D0716204、D0787881zh_TW
dc.identifier.urihttp://dspace.fcu.edu.tw/handle/2377/4659-
dc.description.abstract中文摘要 現今外送服務平台興起,儼然成為一種主流,然而許多校園仍然禁止校外汽、機車進入校園中,導致訂餐者須親自到校門口取餐。因此,本研究目的為建構人工智慧之校園無人車派遣,並研擬相關物流策略,解決目前校園餐飲配送之困境。具體而言,本研究發展校園無人車之AI強化學習餐飲派遣方式,透過基本測試、敏感度分析與情境分析,探究不同物流派遣策略之影響。AI校園無人車之分析結果顯示,以強化學習派遣之求解時間平均小於一秒,在派遣上具有可行性與有效性。物流策略分析結果指出多溫層無人車派遣策略較單一溫層佳,校園規模影響無人車配置規劃。進而,本研究藉由強化學習派遣分析,獲悉不同規模校園特性適宜之餐飲發車點數目、基本派遣車輛數、無人車容量、無人車類型。本研究成果在學術上可作為校園無人車強化學習相關研究之創新發展,實務上可解決餐飲外送無法進入校園之問題。zh_TW
dc.description.abstractAbstract Nowadays, with the rise of the delivery service platform, it has become a mainstream. However, many campuses still prohibit off-campus vehicles and motorcycles from entering the campus, which leads to the fact that the diners have to pick up their meals in person at the school gate. Therefore, the purpose of this study is to build an artificial intelligence campus unmanned vehicle dispatch, and develop related logistics strategies to solve the current dilemma of catering distribution in campus. Specifically, this research develops AI reinforcement learning catering dispatch mode of campus unmanned vehicles, and explores the influence from different logistics dispatch strategies through basic test, sensitivity analysis and scenario analysis. The results from analysis of AI campus unmanned vehicles show that the solution time of the reinforcement learning dispatch is less than one second on average, which is feasible and effective in dispatch. The results of logistics strategy analysis show that the dispatching strategy of multi-temperature unmanned vehicles is better than those of single-temperature, and the campus scale affects the configuration planning of unmanned vehicles. Furthermore, through the reinforcement learning dispatch analysis, the number of catering departure points, the number of basic dispatched vehicles, the capacity of unmanned vehicles, and the type of unmanned vehicles suitable for different sizes of campus were obtained. The results of this research can be used as the innovation and development of the related research on campus unmanned vehicle reinforcement learning, and can solve the problem that the catering delivery can not enter the campus in practice.zh_TW
dc.description.tableofcontents目錄 第一章、前言 1 1.1研究動機 1 1.2 研究目的 1 1.3 研究範圍 2 1.4 研究步驟 3 第二章、文獻回顧 4 2.1車輛路徑問題 4 2.2強化學習 5 2.3強化學習相關模式 10 2.3.1貝爾曼方程(Bellman Equation) 10 2.3.2 Q學習(Q-Learning) 11 2.3.3 策略梯度 13 2.3.4 深度Q網路(Deep Q Network , DQN) 14 2.3.5 Sarsa 16 2.4強化學習應用旅行商問題與車輛路徑問題 17 2.5綜合評析 22 第三章、研究方法 23 3.1 問題特性 23 3.2校園無人車派遣之強化學習架構 23 第四章、結果分析與討論 28 4.1基本測試與分析 28 4.1.1 測試說明 28 4.1.2 測試結果 29 4.1.3 敏感度分析 30 4.1.4 綜合討論 32 4.2情境分析 33 4.2.1 情境參數說明 33 4.2.2 分析測試 33 4.2.3綜合情境分析 44 第五章、管理意涵 48 第六章、結論與建議 51 參考文獻 52zh_TW
dc.format.extent59p.zh_TW
dc.language.isozhzh_TW
dc.rightsopenbrowsezh_TW
dc.subject校園無人車zh_TW
dc.subject餐飲配送zh_TW
dc.subject強化學習zh_TW
dc.subject物流策略zh_TW
dc.subject車輛派遣zh_TW
dc.subjectcampus unmanned vehiclezh_TW
dc.subjectcatering distributionzh_TW
dc.subjectreinforcement learningzh_TW
dc.subjectlogistics strategyzh_TW
dc.subjectvehicle dispatchzh_TW
dc.title校園無人車之AI強化學習餐飲派遣分析zh_TW
dc.title.alternativeAnalysis of AI Reinforcement Learning Catering Dispatch for Unmanned Vehicles on Campuszh_TW
dc.typeUndergracasezh_TW
dc.description.course專題研究zh_TW
dc.contributor.department運輸與物流學系, 建設學院zh_TW
dc.description.instructor吳沛儒-
dc.description.programme運輸與物流學系, 建設學院zh_TW
分類:建109學年度

文件中的檔案:
檔案 描述 大小格式 
D0716278109237.pdf5.7 MBAdobe PDF檢視/開啟


在 DSpace 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。