基於使用者的無樁式公共自行車調度系統

A User-Based Relocation System for Free Floating Bike Sharing System

10.6342/NTU201901759

劉祝銘

無樁式公共自行車系統 ； 基於使用者的車輛調度 ； 分區聚類 ； 需求預測 ； Free Floating Bike Sharing System ； User-based Relocation ； Partition clustering ； Demand prediction

國立臺灣大學土木工程學系學位論文

2019年

碩士

張學孔

繁體中文

隨著近年來人們環境保護意識的覺醒，以及系統租借車輛便利性與經濟性的提升，越來越多的人開始選擇使用公共自行車作為城市内中短程旅次的交通方式，或者以其作為其他公共運具的接駁系統。而公共自行車系統自1965年發展至今，為解決第三代有樁系統中借還站點位置及站點容量帶來的不便，「第四代無樁式公共自行車系統」(Free Floating Bike Sharing System)於2015年在北京誕生，並於近年在中國大陸地區呈現爆炸式的發展。 然而在這快速發展的過程中，一些問題漸漸浮現。例如無樁式公共自行車系統幾乎允許使用者隨處借還車輛因而導致其車輛分佈更為分散無序，再加之每日通勤旅次導致的潮汐性交通流以及部分地點自身具有的起點型或終點型的特性，最終導致車輛在部分時段過於集中而使得周轉率下降或車輛在空間上分佈不均衡而致使系統的服務水準偏低。因此，一套因應無樁式公共自行車相關問題的車輛調度系統是必要的。 既有的公共自行車調度系統多是以營運者駕駛卡車運載自行車到達缺少車輛的借還站點，然而無樁式系統的車輛不似有樁式系統的車輛皆集中於站點，若還僅僅使用基於營運者的調度，其經濟性偏低。針對此現狀，本研究嘗試透過給予公共自行車使用者一個激勵性的獎金，「誘導」使用者將車輛歸還到缺少車輛或將要缺少車輛的地區，以此緩解車輛分佈不均衡的現象。 為構建上述調度系統，本研究首先以目前中國最大無樁式公共自行車企業之一——摩拜單車在北京市的歷史騎行資料，來分析該公共自行車系統使用上的時空特性，進而量化車輛使用在地理上的不均衡；為瞭解使用者對激勵方案的反應模式，本研究使用在線問卷調查以得到使用者的參與意願；接著為達到對車輛的分區管理，本研究使用K-means分區聚類方法，基於車輛使用特性，將研究區域劃分為不同的O-D交通小區；進而使用倒傳遞類神經網絡，通過對歷史O-D量資料的學習，為每個交通小區訓練其近期O-D量預測模式；基於預測得到的各小區下時間段O-D量，即可定義各交通小區可供調度的車輛數及存車缺口量，本調度系統以最小調度成本為目標，同時限制存車缺口，進行車輛調度規劃。本研究選擇了3個不同類型的區域模擬應用本調度系統48小時以進行案例分析，結果顯示本調度系統在各個區域皆以可承受的調度成本獲得了顯著的改善，證明了系統的實用性，而調度效果會因地區而異，研究建立之調度系統及分析方法，可以提供對無樁式公共自行車相關業者參考借鑒並對其車輛調度策略進行優化。

In recent years, with people's increased awareness of protecting the environment, more and more people choose to use public bike travel within city or as a connecting tool for other public transit modes. Since the first generation public bicycle system has been developed in 1965, the fourth generation which called Free Floating Bike Sharing (FFBS) System has been developed and exploded in Beijing in 2015 to solve the problems caused by the location and station capacity of the third generation station-based system. The emergence of so many bikes in a short time recent years can cause many problems. For instance, the free floating system allows the user to rent and return the bike almost from anywhere within the operating area, thus resulting in an imbalance distribution of bikes. In addition, the tidal flow caused by daily commuting trip can cause bikes to be concentrated and lower the turnover rate, and there have some origin or destination type regions will cause imbalance distribution of bikes as well and resulting in a low service level; therefore, a relocation system is necessary. The existing operator-based relocation system for public bike system mostly uses the operator to drive a truck to carry the bicycle to those insufficient stations. However, the characteristics of free floating makes if we only use the traditional operator-based relocation strategy, it will become economically less attractive. In response to this situation, this research attempts to alleviate the imbalance distribution of bikes by giving users an incentive bonus to “employ” users to return the bike to the near area that lack or will lack bikes. In order to construct the above-mentioned dispatching system, based on the historical trip data of Mobike which is one of the biggest free floating bike sharing company in China, this study first analyzed the temporal and spatial characteristics of the bike usage to quantify the imbalance distribution. Then, based on GPS data, spatial clustering was used to construct transit Origin-Destination matrix. Additionally, in order to predict upcoming Origin-Destination traffic volume at certain district, depending on different temporal factors, a demand model was built using back-propagation neural network (BPNN). The prediction result was used to compute the redundant bike that could be dispatched and the insufficiency at each district. Then, combined with a user participation pattern that established based on an online survey, a dispatching system that considers minimizing both the incentives payout and the total bike shortfall developed and evaluated. In the end, a case study of free floating bike sharing system was performed to show the effect of the proposed system within 3 different types of case regions. Although the result shows that the relocation effect varies by region. However, the relocation system in each region can achieve significant improvement under an affordable dispatching cost. The result proves the practicality of the user-based relocation system, and this research is also a valuable reference for the free floating bike sharing related company to optimize their bike relocation strategies.

1. 英文文獻：
2. 1. Banerjee, A. & Dave, R. N. (2004). Validating clusters using the Hopkins statistic. 2004 IEEE International Conference on Fuzzy Systems, IEEE Cat. No. 04CH37542, 1, pp. 149-153.
3. 2. Bishop, R. C. & Heberlein, T. A. (1979). Measuring values of extra market goods: Are indirect measures biased? American journal of agricultural economics, 61(5), pp. 926-930.
4. 3. Chemla, D. Meunier, F. & Calvo, R. W. (2013). Bike sharing systems: Solving the static rebalancing problem. Discrete Optimization, 10(2), pp. 120-146.
5. 4. Davies, D. L. & Bouldin, D. W. (1979). A cluster separation measure. IEEE transactions on pattern analysis and machine intelligence, (2), pp. 224-227.
6. 5. Davis, L. S. (2014). Rolling along the last mile: Bike-sharing programs blossom nationwide. Planning, 80(5), pp. 10-16.
7. 6. DeMaio, P. (2009). Bike-sharing: History, impacts, models of provision, and future. Journal of public transportation, 12(4), pp. 41-56.
8. 7. Fishman, E. Washington, S. Haworth, N. & Mazzei, A. (2014). Barriers to bikesharing: an analysis from Melbourne and Brisbane. Journal of Transport Geography, 41, pp. 325-337.
9. 8. Hanemann, M. Loomis, J. & Kanninen, B. (1991). Statistical efficiency of double-bounded dichotomous choice contingent valuation. American journal of agricultural economics, 73(4), pp. 1255-1263.
10. 9. ITDP (2018). The Bikeshare Planning Guide. Institute for Transportation and Development Policy, 2018.
11. 10. Kwok, T. Y. & Yeung, D. Y. (1997). Constructive algorithms for structure learning in feedforward neural networks for regression problems. IEEE transactions on neural networks, 8(3), pp. 630-645.
12. 11. Li, Y. Lu, J. Zhang, L. & Zhao, Y. (2017). Taxi booking mobile app order demand prediction based on short-term traffic forecasting. Transportation Research Record, 2634(1), pp. 57-68.
13. 12. Luo, D. Cats, O. & van Lint, H. (2017). Constructing transit origin–destination matrices with spatial clustering. Transportation Research Record, 2652(1), pp. 39-49.
14. 13. Pal, A. & Zhang, Y. (2017). Free-floating bike sharing: Solving real-life large-scale static rebalancing problems. Transportation Research Part C: Emerging Technologies, 80, pp. 92-116.
15. 14. Peking University (2015). “ofo bicycle”: riding bicycles anytime and anywhere,
16. from: https://web.archive.org/web/20171028145810/http://english.pku.edu.cn/news_events/news/campus/4059.htm.
17. 15. Pfrommer, J. Warrington, J. Schildbach, G. & Morari, M. (2014). Dynamic vehicle redistribution and online price incentives in shared mobility systems. IEEE Transactions on Intelligent Transportation Systems, 15(4), pp. 1567-1578.
18. 16. Qiu, L. Zhang, D. Huang, H. Xiong, Q. & Zhang, G. (2018). BP Neural Network Based Prediction of Potential Mikania micrantha Distribution in Guangzhou City. Forest Res, 7(216), 2, pp. 1-6.
19. 17. Reiss, S. & Bogenberger, K. (2017). A Relocation Strategy for Munich's Bike Sharing System: Combining an operator-based and a user-based Scheme. Transportation Research Procedia, 22, pp.105-114.
20. 18. Singla, A. Santoni, M. Bartók, G. Mukerji, P. Meenen, M. & Krause, A. (2015). Incentivizing users for balancing bike sharing systems. Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence, pp. 723-729.
21. 19. Vlahogianni, E. I. & Karlaftis, M. G. (2013). Testing and comparing neural network and statistical approaches for predicting transportation time series. Transportation Research Record, 2399(1), pp. 9-22.
22. 20. Vogel, P. Greiser, T. & Mattfeld, D. C. (2011). Understanding bike-sharing systems using data mining: Exploring activity patterns. Procedia-Social and Behavioral Sciences, 20, pp. 514-523.
23.  
24. 中文文獻：
25. 1. Mobike cup (2017)，2017 摩拜杯演算法挑戰賽,
26. 取自：https://biendata.com/competition/mobike/.
27. 2. 丁靜 (2017年)，北京改善綠色出行環境：兩年治理468公里步行自行車系統，新華網。
28. 3. 中國信息通信研究院政策與經濟研究所、摩拜單車 (2018)，中國共享單車行業發展報告(2018)。
29. 4. 中華人民共和國國家統計局 (2011)，第六次全國人口普查資料，
30. 取自：https://www.citypopulation.de/php/china-township-beijing-admin_c.php.
31. 5. 國家信息中心分享經濟研究中心 (2017)，共享單車行業就業研究報告。
32. 6. 比達網 (2017)，2017年第1季度中國共享單車市場研究報告。
33. 7. 北京市交通委員會 (2017年)，北京市鼓勵規範發展共享自行車的指導意見(試行)。
34. 8. 北京市統計局、國家統計局北京調查總隊 (2018)，北京統計年鑒2018. pp. 11。
35. 9. 北京清華同衡規劃設計研究院、摩拜單車 (2017)，2017年共享單車與城市發展白皮書。
36. 10. 交通運輸部 (2017)，交通運輸部等10部門關於鼓勵和規範互聯網租賃自行車發展的指導意見。
37. 11. 艾媒諮詢 (2017)，2017Q1中國共享單車市場研究報告。
38. 12. 呂玉強、秦勇、賈利民、董宏輝、賈獻博與孫智源 (2010)，基於計程車GPS資料聚類分析的交通社區動態劃分方法研究. 物流技術, 216, pp. 86-88。
39. 13. 翁偉倫 (2018)，公共自行車使用者於他站租還車之願受價格研究-以台北市YouBike為例，國立交通大學運輸與物流管理學系碩士論文。
40. 14. 張斐章、張麗秋 (2005)，類神經網路，東華書局。
41. 15. 張學孔、賴勁丞、邱弈珩 (2016)，基於站點相依性之公共自行車調度策略，中華民國運輸學會105年學術論文研討會。
42. 16. 郭瑞雪 (2017)，基於BP神經網路的網約車出行需求短時預測，北京交通大學交通運輸學院碩士論文。
43. 17. 廖應成 (2017)，共享單車電子圍欄技術使用調研報告，RFID世界網。
44. 18. 潘敬文 (2019)，摩拜廣州上線近百個電子圍欄禁停區，信息時報。
45. 19. 羅子瑛 (2017)，北京試點共享單車電子圍欄北斗導航定位，新華網。