Muroran-IT Academic Resource Archive
Not a member yet
5947 research outputs found
Sort by
機械学習技術の視点による通勤交通手段選択と自動車所有に関する研究
Get PDF室蘭工業大学Muroran Institute of Technology博士（工学）application/pdf過去数十年の間に，世界は急速な都市化プロセスを経験し，人々の生活には自動車が急速に普及した。モータリゼーションは我々に経済発展の機会を与えると同時に，生活の質に影響を与える地球環境に負荷をかけている。都市の自己増殖は，自家用車の所有と使用の増加を引き起こす主要な理由である。旅行モードの選択，車両所有パターン，およびそれらの決定要因に対して影響力のあるメカニズムを理解することは，土地利用と交通計画上の政策決定に大いに役立つ。この課題は，グローバリゼーションの時代に持続可能な交通の発展を目指す途上国において，大いに注目されている。本研究では，多項ロジットモデル，ニューラルネットワーク，ランダムフォレストを用いて，プノンペン市における将来のインパクトレベルと車両所有パターンの予測を行った。交通手段選択に関して，本研究では，勾配ブースティングマシン，機械学習アルゴリズム，およびLIMEを適用して，インドネシアのジャカルタ市の大都市圏における複数交通手段によるトリップパターンとその決定要因を推定した。両方の分析は，国際協力機構（JICA）から提供された世帯インタビュー調査データを使用した。分析結果は，家計収入がプノンペンのモータリゼーションに影響を与える最も強力な変数であることを示した。合計旅行回数などの個々の旅行特性の補足，通勤目的で行われた移動回数と全体の移動距離は全て，分類子として効果的に作用した。ジャカルタ市におけるケーススタディにおいては，単一交通機関の旅行に影響を与える要因として旅行費用や移動時間といった限られた変数が選ばれる一方で，複数交通機関の旅行については幅広い変数の影響を受けていることが示された。さらに，機械学習アプローチによる予測においては，精度を予測するという点だけでなく，統計的アプローチと比較して不均衡なカテゴリを処理するという点でも優れていたことが認められた。特に，グラディエントブースティングマシンは，ビッグデータで課題を解決する際，優れた潜在能力があることが示された。これら二つの結果は，旅行行動分析の分野に関して機械学習技術を適用する優位性を示し，他の分析に関しても，機械学習技術が応用できる可能性を示唆している。In the last decades, the world has seen the rapid urbanization process with the boom of motorized vehicles. The motorization, on one hand, gives opportunities for economic development and on the other hand, it puts pressure on the environment that affects the quality of life. The self-proliferating of the city is identified as a major that causes the rise of private vehicle ownership and usage. Understanding the influential mechanism of the travel mode choice, vehicle ownership patterns, and their determinants will greatly help policymaking for land use and transportation. This issue has been paid even greater attention in developing countries that aspire to reach sustainable transportation development goals in the era of globalization.In this study, the Multinomial logit model, Neural Networks and Random Forest were applied to examine the features’ impact level and to also predict vehicle ownership patterns in Phnom Penh city. Regarding travel mode choice, this study introduces the application of Gradient Boosting Machine, a Machine Learning algorithm, and Local Interpretable Model-agnostic Explanations technique to investigate the multi-mode trip pattern and its determinants in the metropolitan area of Jakarta city, Indonesia. Both analyses used the household interview survey data provided by the Japan International Cooperation Agency (JICA). The results indicate that household income is the most powerful variable affecting motorization in Phnom Penh. Supplementation of individual trip characteristics such as total number of trips made, number of trips made for work purposes and overall travel distance all make effective contributions as classifiers. The results from the case study of Jakarta city show that there was a limit of features (travel cost, time, etc.) that affected the single-mode trip while the multi-mode travel was influenced by the wide range of variables. Furthermore, it is acknowledged that the machine-learning approach outperformed not only in terms of predicting accuracy but also in dealing with unbalanced categories when compared with the statistical approach. Especially, the Gradient Boosting Machine indicated the impressive potentiality in solving the subject with big data. This detection supplies the advantages of applying machine learning techniques in terms of, but not limited to, the field of travel behavior.doctoral thesi
機械学習による都市公園の利用についての誘発要因に関する推論
Get PDFapplication/pdf本研究は，都市公園の利用を誘発する要因を機械学習により推定することを目的とした。調査は，東京都江東区に在住する地域住民を対象に都市公園の利用状況や自身の生活状況に関するオンラインアンケート調査を実施し，最終的に1,553 名から回答を得た。はじめに，潜在クラス分析により都市公園における利用行動の実施状況を低利用型および利用型の2クラスターに分類し，利用行動クラスターとした。続いて，決定木分析の目的変数として利用行動クラスターを設定し，その他の全調査項目を説明変数として予測モデルを構築した。この結果，都市公園の利用を誘発する要因として，“緑地における家族との交流”，“緑道の利用頻度”などが推定された。journal articl
迅速な災害管理のための即時的,持続可能，かつ拡張的なエッジコンピューティングの研究
Get PDF室蘭工業大学Muroran Institute of Technology博士（工学）application/pdf本学位論文は、迅速な災害管理におけるいくつかの問題に取り組んだ。既存のネットワークインフラが災害による直接的なダメージや停電によって使えないことを想定し、本論文では、最新のICTを用いた次世代災害支援システムの構築を目指す。以下のとおり本論文は三部で構成される。第一部は、災害発生後の緊急ネットワーキングである。本論文では、情報指向フォグコンピューティング（Information-Centric Fog Computing）というアーキテクチャを提案し、既存のインフラがダウンした場合に臨時的なネットワーク接続を提供する。本論文では、六次の隔たり理論から着想を得て、緊急時向け名前ベースルーティング（Name-Based Routing）を考慮した。まず、二層の情報指向フォグコンピューティングネットワークモデルを提案した。次に、ソーシャルネットワークを元に、情報指向フォグノード間の関係をモデリングし、名前ベースルーティングプロトコルをデザインする。シミュレーション実験では、既存のソリューションと比較し、提案手法はより高い性能を示し、有用性が証明された。第二部は、ネットワークの通信効率の最適化である。本論文は、第一部で構築されたネットワークの通信効率を最適化し、ネットワークの持続時間を延ばすために、ネットワークのエッジで行われるキャッシングストラテジーを提案した。本論文では、まず、第一部で提案した二層ネットワークモデルをベースにサーバー層も加えて、異種ネットワークストラクチャーを構成した。次に、緊急時向けのエッジキャッシングに必要なTime to Live （TTL）とキャッシュ置換ポリシーを設計する。シミュレーション実験では、エネルギー消費とバックホールレートを性能指標とし、メモリ内キャッシュとディスクキャッシュの性能を比較した。結果では、メモリ内ストレージと処理がエッジキャッシングのエネルギーを節約し、かなりのワークロードを共有できることが示された。第三部は、ネットワークカバレッジの拡大である。本論文は、ドローンの関連技術とリアルタイム視覚認識技術を利用し、被災地のユーザ捜索とドローンの空中ナビゲーションを行う。災害管理におけるドローン制御に関する研究を調査し、現在のドローン技術と無人捜索救助に対する実際のニーズを考慮すると、軽量なソリューションが緊急時に必要であることが判明した。そのため本論文では、転移学習を利用し、ドローンに搭載されたオンボードコンピュータで実行可能な空中ビジョンに基づいたナビゲーションアプローチを開発した。シミュレーション実験では、1/150ミニチュアモデルを用いて、空中ナビゲーションの実行可能性をテストした。結果では、本論文で提案するドローンの軽量ナビゲーションはフィードバックに基づいてリアルタイムに飛行の微調整を実現でき、既存手法と比較して性能において大きな進歩を示した。This dissertation mainly focuses on solving the problems in agile disaster management. To face the situation when the original network infrastructure no longer works because of disaster damage or power outage, I come up with the idea of introducing different emerging technologies in building a next-generation disaster response system. There are three parts of my research. In the first part of emergency networking, I design an information-centric fog computing architecture to fast build a temporary emergency network while the original ones can not be used. I focus on solving name-based routing for disaster relief by applying the idea from six degrees of separation theory. I first put forward a 2-tier information-centric fog network architecture under the scenario of post-disaster. Then I model the relationships among ICN nodes based on delivered files and propose a name-based routing strategy to enable fast networking and emergency communication. I compare with DNRP under the same experimental settings and prove that my strategy can achieve higher work performance. In the second part of efficiency optimization, I introduce the idea of edge caching in prolong the lifetime of the rebuilt network. I focus on how to improve the energy efficiency of edge caching using in-memory storage and processing. Here I build a 3-tier heterogeneous network structure and propose two edge caching methods using different TTL designs & cache replacement policies. I use total energy consumption and backhaul rate as the two metrics to test the performance of the in-memory caching method and compare it with the conventional method based on disk storage. The simulation results show that in-memory storage and processing can help save more energy in edge caching and share a considerable workload in percentage. In the third part of coverage expansion, I apply UAV technology and real-time image recognition in user search and autonomous navigation. I focus on the problem of designing a navigation strategy based on the airborne vision for UAV disaster relief. After the survey of related works on UAV fly control in disaster management, I find that in consideration of the current UAV manufacturing technology and actual demand on unmanned search & rescue, a lightweight solution is in urgent need. As a result, I design a lightweight navigation strategy based on visual recognition using transfer learning. In the simulation, I evaluate my solutions using 1/150 miniature models and test the feasibility of the navigation strategy. The results show that my design on visual recognition has the potential for a breakthrough in performance and the idea of UAV lightweight navigation can realize real-time flight adjustment based on feedback.doctoral thesi
Three-dimensional real time imaging of amyloid β aggregation on living cells
Get PDFapplication/pdfAlzheimer’s disease (AD) is a progressive disorder of the brain that gradually decreases thinking, memory, and language abilities. The aggregation process of amyloid β (Aβ) is a key step in the expression of its neurocytotoxicity and development of AD because Aβ aggregation and accumulation around neuronal cells induces cell death. However, the molecular mechanism underlying the neurocytotoxicity and cell death by Aβ aggregation has not been clearly elucidated. In this study, we successfully visualized real-time process of Aβ42 aggregation around living cells by applying our established QD imaging method. 3D observations using confocal laser microscopy revealed that Aβ42 preferentially started to aggregate at the region where membrane protrusions frequently formed. Furthermore, we found that inhibition of actin polymerization using cytochalasin D reduced aggregation of Aβ42 on the cell surface. These results indicate that actin polymerization-dependent cell motility is responsible for the promotion of Aβ42 aggregation at the cell periphery. 3D observation also revealed that the aggregates around the cell remained in that location even if cell death occurred, implying that amyloid plaques found in the AD brain grew from the debris of dead cells that accumulated Aβ42 aggregates.journal articl
Response to Hydrogen in Field-effect Transistor Sensor with Platinum Nanoparticles
Get PDFapplication/pdfA novel device based on nanometer size platinum (Pt) on the gate electrode of metal-oxidesemiconductor field-effect transistor (MOSFET) for detecting hydrogen (H2) gas was fabricated. The operation characteristics of the device for the detection of H2 gas presents as a function of H2 gas concentration. The drain current in the output characteristics of the MOSFET increased rapidly depending on the H2 gas concentration. It was possible to detect 1,000 ppm for H2 gas at room temperature. A model was proposed to explain the operation. The sensing mechanism of the device is supported well by the experimental data.学術論文departmental bulletin pape
Slope failures over a wide area in the 2018 Hokkaido Eastern Iburi Earthquake and its evaluation
Get PDFapplication/pdfThe 2018 Hokkaido Eastern Iburi Earthquake caused a large amount of slope failures over a wide area.In particular, the enormous damage due to slope failures was centered in Atsuma and Abira town which are to the north of the epicenter. In Atsuma town, a seismic intensity of seven has been observed for the first time in Hokkaido, Japan. The geology in this region was mainly formed from three kinds of pyroclastic fall deposits (fa) erupted from Tarumae volcano, Eniwa volcano, and Shikotsu Caldera.Especially, the serious damage was generated in these tephra stratus. This report summarizes earthquake-induced damages on natural slopes, and presents physical feature and mechanical properties of the collapsed pyroclastic fall deposits distributed over this area.特集departmental bulletin pape
Survey on Factors of Geotechnical Disaster due to 2018 Hokkaido Eastern Iburi earthquake
Get PDFapplication/pdfNatural disasters caused by huge earthquake have occurred in various places, and has made a great impact on social activities and people's lives in recent years in japan. Then the effect on transportation caused by disaster on earthquake is very high. The Hokkaido Eastern Iburi earthquake in 2018 caused geotechnical disasters, which are large slope failures in wide area, and unique liquefaction at Satozuka area in Sapporo city etc. In particular, serious geotechnical damages were induced in Atsuma town recorded 7 on the Japanese seismic intensity scale (a magnitude of 6.7). The objective of this study is to investigate factors for geotechnical disaster due to 2018 Hokkaido Eastern Iburi earthquake in these region. In this study, geomorphic and geological investigation for factor of slope failure occurred in the Atsuma town and that of liquefaction at Satozuka area in Sapporo city is discussed. In the survey, it is found that the large slope failure occurred in the Atsuma town is caused by Ta-d unit of Tarumae volcanic ash layer, and the type of slope failure is surface failure.特集departmental bulletin pape
Distribution Map of Frost Resistance for Cement-Based Materials Based on Pore Structure Change
Get PDFapplication/pdfThis paper presents a prediction method and mathematical model based on experimental results for the change in pore structure of cement-based materials due to environmental conditions. It focuses on frost damage risk to cement-based materials such as mortar. Mortar specimens are prepared using water, ordinary Portland cement, and sand and the pore structure is evaluated using mercury intrusion porosimetry. New formulas are proposed to describe the relationship between the pore structure change and the modified maturity and to predict the durability factor. A quantitative prediction model is established from a modified maturity function considering the influences of environmental factors like temperature and relative humidity. With this model, the frost resistance of cement-based materials can be predicted based on weather data. Using the prediction model and climate data, a new distribution map of frost damage risk is created. It is found that summer weather significantly affects frost resistance, owing to the change in pore structure of cement-based mortar. The model provides a valuable tool for predicting frost damage risk based on weather data and is significant for further research.journal articl
Secure and Efficient Vehicle-to-Grid Energy Trading in Cyber Physical Systems: An Integration of Blockchain and Edge Computing
Get PDFapplication/pdfSmart grid has emerged as a successful application of cyber-physical systems in the energy sector. Among numerous key technologies of the smart grid, vehicle-to-grid (V2G) provides a promising solution to reduce the level of demand-supply mismatch by leveraging the bidirectional energy-trading capabilities of electric vehicles. In this paper, we propose a secure and efficient V2G energy trading framework by exploring blockchain, contract theory, and edge computing. First, we develop a consortium blockchain-based secure energy trading mechanism for V2G. Then, we consider the information asymmetry scenario, and propose an efficient incentive mechanism based on contract theory. The social welfare optimization problem falls into the category of difference of convex programming and is solved by using the iterative convex-concave procedure algorithm. Next, edge computing has been incorporated to improve the successful probability of block creation. The computational resource allocation problem is modeled as a two-stage: 1) Stackelberg leader-follower game and 2) the optimal strategies are obtained by using the backward induction approach. Finally, the performance of the proposed framework is validated via numerical results and theoretical analysis.journal articl
高圧下で合成された充填スクッテルダイト化合物 Smy(Fe1-xNix)4Sb12 の熱電特性
Get PDF室蘭工業大学Muroran Institute of Technology博士（工学）application/pdf熱電変換材料は，効率的に熱エネルギーを電気エネルギーに直接変換可能な材料であり，熱電発電や熱電冷却に不可欠な電子材料である。熱電変換材料の性能は，性能指数Z＝S2/ (S：ゼーベック係数，：電気抵抗率，：熱伝導率)で評価され，これに絶対温度Tを乗じた無次元性能指数ZT≧1 が実用化の目安となっている。熱電変換モジュールには，n型とp型の熱電変換材料が必要であるが，同じ結晶構造，同じ構成元素からn型，p型両方の材料が合成できる材料系は，両者の熱膨張係数が同程度であるため，モジュールを作製する場合に有利である。Sbを含む充填スクッテルダイト化合物RT4Sb12 は，優れた熱電特性を有することから，次世代の高性能熱電変換材料への応用が期待されている物質である。さらに，Rの充填率やT元素の置換により，n型，p型両方の材料合成が可能である。最近，充填スクッテルダイト化合物Smy(Fe1-xNix)4Sb12において， Smの充填量yとFe, Niの比率xを変えることで，p型からn型へと変化することが示され，比較的高いZTが報告された。しかし，従来の合成方法では，x，yの可変量には限界があり，広い範囲でSmy(Fe1-xNix)4Sb12の熱電特性を調べることはできなかった。そこで，我々は高圧合成法により，この問題を解決し，この系におけるZTのさらなる向上を目指した。本研究では，高圧合成法により，Smの充填量y，Fe/Niの比xを広い範囲で変化させた充填スクッテルダイト化合物Smy(Fe1-xNix)4Sb12の試料合成を試み，試料評価，熱電特性評価を行った。従来の合成方法では，Ni置換量x = 0.4付近で，実際のSmの充填量は0.3以下であったのに対し，同様のNi置換量で，Smの最大充填量はy = 0.56と大きく向上した。このことは，高圧合成法により，Ni置換量を変化させずにSmの充填量を大きく変化せることができることを示しており，この系の設計の幅を広げることにつながると考えられる。また，ゼーベック係数，ホール効果の測定から，Smの充填率yとFe, Ni の比xを変えることで，p型伝導体，ｎ型伝導体を作り分けることに成功した。室温におけるn型Smy(Fe1-xNix)4Sb12 のZT の最大値は0.167であり，高温ではさらに高いZTを示す可能性がある。これらの結果は，今後の熱電変換材料の設計の幅を広げることにつながり，次世代の高性能な熱電変換材料開発の発展に寄与ものであると考えられる。The efficiency of thermoelectric (TE) materials is determined by the dimensionless figure of merit ZT=S2T/ρκ, where S is the Seebeck coefficient, T is the absolute temperature, ρ is the electrical resistivity, and κ is the thermal conductivity. TE devices require both n-type and p-type materials. It is important that both of them can be obtained from the same matrix with the same crystal structure because small mismatch in the thermal expansion coefficient is important for fabricating a TE module. Sb-based filled skutterudite compounds RT4Sb12 have attracted considerable attention as one of the best candidates of TE materials. Furthermore, both p-type and n-type filled skutterudite compounds can be synthesized depending on the filling ratio of R atom and the partial substitution of T atom. Recent studies of Smy(Fe1-xNix)4Sb12 indicated the p/n crossover and relatively high ZT value. In this system, the TE properties can be tuned by the Sm filling ratio y and the substitution Fe/Ni content x. However, the tunable range of the filling ratio y of Sm atom and Fe/Ni substitution ratio x is limited by conventional synthesis methods. High-pressure synthesis (HPS) using multi-anvil press could overcome this problem. This method enables to increase filling fraction of R atom in the filled skutterudite structure in comparison to conventional methods. The optimized ZT in Smy(Fe1-xNix)4Sb12 could be achieved by carefully expanding the research areas of synthesis with tuning x and y. In this study, we have tried to synthesize Smy(Fe1-xNix)4Sb12 using the HPS method. The structure and chemical composition of the samples were studied. The maximum value of actual Sm filling ratio was y = 0.56 with actual Ni content x ~ 0.40 while y < 0.3 in samples with similar x values prepared by conventional methods. By using HPS technique, we could prepare Smy(Fe1-xNix)4Sb12 samples with various Sm filling ratio in spite of almost the same Ni content. The p/n crossover determined by Seebeck and Hall coefficients was around x = 0.3 of Ni content. P-type and n-type skutterudites could be obtained by tuning Sm-filling ratio and Fe/Ni content. The highest ZT value of 0.167 was achieved for n-type skutterudite at room temperature. HPS using multi-anvil press made it possible to enlarge the variation of TE material design. These results contribute development of next-generation TE materials.doctoral thesi