Fast Railway Delay Evaluation Method Based on Discrete Distribution Propagation

• Item type: SIG Technical Reports(1)
• 公開日: 2019-02-21
• タイトル: Fast Railway Delay Evaluation Method Based on Discrete Distribution Propagation
• タイトル / 言語: en
• タイトル: Fast Railway Delay Evaluation Method Based on Discrete Distribution Propagation
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Toshiba Corporation
• 著者所属: Toshiba Corporation
• 著者所属: Toshiba Digital Solutions Corporation
• 著者所属(英): en
• 著者所属(英): Toshiba Corporation
• 著者所属(英): en
• 著者所属(英): Toshiba Corporation
• 著者所属(英): en
• 著者所属(英): Toshiba Digital Solutions Corporation
• 著者名: Tomoshi, Otsuki; Hideo, Sakamoto; Hideki, Kubo
• 著者名(英): Tomoshi, Otsuki; Hideo, Sakamoto; Hideki, Kubo
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Accurate delay evaluation of timetables is crucial for railway companies. There have been some conventional methods utilizing continuous random variables directly. These methods, however, suffer combinatorial expansion problems, which require complex pruning techniques. In this paper, discretizing the delay distribution on railway networks, we present a method calculating propagated delay distributions on each event analytically under the assumption of propagated delays' independence. We also show the complexity of the proposed method is O (M²N) in the general case and can be reduced to O (MN) especially for the special cases where the source distribution is the negative binominal distribution, where M denotes the number of quantization levels in discretization and N denotes the total number of events. Finally, computational experiments on test scenarios with M ～ 700 show that the proposed method provides almost same results as and in over 500 times faster computation time than Monte Carlo simulations.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Accurate delay evaluation of timetables is crucial for railway companies. There have been some conventional methods utilizing continuous random variables directly. These methods, however, suffer combinatorial expansion problems, which require complex pruning techniques. In this paper, discretizing the delay distribution on railway networks, we present a method calculating propagated delay distributions on each event analytically under the assumption of propagated delays' independence. We also show the complexity of the proposed method is O (M²N) in the general case and can be reduced to O (MN) especially for the special cases where the source distribution is the negative binominal distribution, where M denotes the number of quantization levels in discretization and N denotes the total number of events. Finally, computational experiments on test scenarios with M ～ 700 show that the proposed method provides almost same results as and in over 500 times faster computation time than Monte Carlo simulations.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AN10505667
• 書誌情報: 研究報告数理モデル化と問題解決（MPS）; 巻 2019-MPS-122, 号 7, p. 1-6, 発行日 2019-02-21
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8833
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会