Muroran-IT Academic Resource Archive
Not a member yet
5947 research outputs found
Sort by
Water injection in underground coal gasification with a horizontal hole: A strategy to prevent steel pipe rupture and enhancing hydrogen production reaction
No full textUnderground Coal Gasification (UCG) is a technology that enables the extraction of coal energy by converting coal seams into syngas, which mainly consists of H2, CO, and CH4. During UCG, the temperature in the reaction zone can exceed 1300 °C, raising concerns regarding the potential melting of the steel pipes used for oxidant injection. To mitigate this issue, this study investigated the use of water as an injection agent. Water injection serves two key purposes: cooling the injection pipe and enhancing H2 production. To examine the effects of water injection on temperature in gasification zone and product gas composition, a UCG model experiment was conducted. The results show that water injection effectively inhibit melting pipe by decreasing the temperature in the gasification zone without compromising H2 production, although the CO concentration decreases and the CO2 concentration increases. Additionally, the energy recovery loss due to water injection can be estimated based on the amount of water injected, as the heat loss from water evaporation is the dominant factor. These findings demonstrate that water injection is a viable strategy for preventing pipe melting while enhancing H2 production in UCG processes.journal articl
実務的なCAD製図を可能とする手書きCADインタフェースの実現
Get PDF室蘭工業大学Muroran Institute of Technology博士（工学）スマートフォンやタブレット端末の普及により，フリーハンドによる直感的なスケッチ入力が注目されている．しかし，既存の手法の多くは芸術的な用途に偏っており，工学的用途への応用はほとんど見られない．既存の手書きCADインタフェースSKITは一貫した手書き入力により幾何作図を可能とするが，実務的なCAD製図には課題が残る．SKITを拡張し，実務的なCAD製図を実現するには「入力の安定性の改善」，「キャンバス操作」，「CADコマンド入力」の3つ要件に対応する必要がある．本研究では，まず，入力の安定化のために，制約付きファジィスプライン生成法と局所制御可能な逐次型ファジィスプライン曲線生成法を提案する．次に，キャンバス操作のためにデュアルファジィストローク同定法を提案する．さらに，手書きCADコマンド入力のためにコマンドストローク/描画ストローク弁別法を提案する．最後に，これらの手法をSKITに統合することで，キーボード操作や数値入力に頼らず直感的な手書き操作のみで実務的なCAD製図を達成可能であることを，作図実験を通して示す．The widespread adoption of smartphones and tablet devices has brought significant attention to intuitive freehand sketch input. However, most existing methods are heavily biased towards artistic applications, with very little focus on engineering applications. While the existing handwriting-based CAD interface, SKIT, allows for geometric drawing through consistent handwritten input, challenges remain in achieving practical CAD drafting. To extend SKIT and enable practical CAD drafting, we need to address three key requirements: "improving input stability," "canvas manipulation," and "CAD command input." This research first proposes a Constrained Fuzzy Spline Interpolation method and a Locally Controllable Sequential Fuzzy Spline Curve Generation method to enhance input stability. Next, we propose a Dual Fuzzy Stroke Identification method for canvas manipulation. Furthermore, to facilitate handwriting-based CAD command input, we propose a Command Stroke/Drawing Stroke Discrimination method. Through drawing experiments, we will demonstrate that by integrating these methods into SKIT, it's possible to achieve practical CAD drafting using only intuitive handwritten input, without relying on keyboard
operations or numerical input.doctoral thesi
単目的問題への分割に基づく多目的分枝限定法の提案とその応用
Get PDF室蘭工業大学Muroran Institute of Technology博士（工学）現代社会において，多目的最適化はコスト・品質や効率・環境負荷など，複数の競合する
評価基準のバランスを取る上で不可欠な手法となっている．特に，意思決定に整数制約が含
まれる多目的混合整数線形計画問題（MOMILP）は，幅広い領域で応用可能である一方，解探
索空間が膨大であるため，効率的かつ厳密な解法の開発が重要な課題である．
従来の進化計算手法は多目的問題に対して計算効率が高いものの，厳密解を保証できな
いという欠点がある．一方，多目的分枝限定法は厳密解を求められるが，計算コストが大き
く，実問題への適用が容易ではなかった．
本論文では，これらの課題を克服するために，分解ベースのアプローチと多目的分枝限定
法を統合した新しい厳密解法「Multi-Objective Branch-and-Bound based on
Decomposition（MOBB/D）」を提案する．本手法では，MOMILP を複数の単目的部分問題に分
割し，重みベクトルに基づく単目的最適化を繰り返すことでパレート最適解集合を構築す
る．特に，シンプレックス表や切除平面などの探索情報を部分問題間で再利用する戦略を導
入し，計算コストを削減しながら高品質なパレートフロントの厳密解を得ることを実現し
た．
提案手法の有効性を検証するため，数値実験では既存の多目的分枝限定法との比較を行
い，計算効率と解の質の両面で優位性を示した．さらに，室蘭市のごみ収集問題を対象とし
た実問題への適用例においても，従来手法と比較して走行距離と住民負担のトレードオフ
を高い精度で探索できることを確認した．これは，行政や住民など複数の利害関係者が協議
の上で政策立案を行う場面にも有用であり，EBPM（Evidence-Based Policy Making）の観点
からも大きな意義を持つ．また，本研究で開発したフレームワークは，重みベクトルの動的
な設計や探索空間の部分的再定式化など，さらなる機能拡張によって汎用性を高められる
可能性がある．これにより，より複雑な制約や大規模な実用問題にも対応でき，多目的最適
化の現場への普及を促進することが期待される．
以上の成果により，MOBB/D は厳密解法としての妥当性と，多目的最適化に求められる実
用的な計算効率を両立する新たな手法として位置づけられる．将来的には，探索情報の取捨
選択や列生成法などのさらなる活用，ならびに非線形計画問題への拡張など，多面的な発展
が期待される．本研究の知見は，多目的最適化分野全般において，新たな方向性を示すとと
もに，実問題に対する高度な意思決定支援の可能性を大きく広げるものである．In modern society, multi-objective optimization plays an indispensable role in
balancing multiple competing objectives, such as cost, quality, efficiency, and
environmental impact. Among these problems, multi-objective mixed-integer linear
programming (MOMILP) is particularly relevant due to its applicability across
various domains, although it poses formidable computational challenges.
Evolutionary computation methods, while effective in handling large-scale and highdimensional
problems, cannot guarantee exact solutions. Conversely, multiobjective
branch-and-bound methods can offer exact Pareto-optimal solutions but
often suffer from high computational cost, making them difficult to apply to realworld
problems.
In this dissertation, we propose a novel approach called Multi-Objective Branchand-
Bound based on Decomposition (MOBB/D), which integrates a decomposition-based
strategy with multi-objective branch-and-bound. Specifically, we decompose an
MOMILP into multiple single-objective subproblems, each parameterized by a weight
vector. By repeatedly solving these subproblems with exact optimization techniques,
we collectively construct the Pareto-optimal solution set. Furthermore, we
introduce a key innovation of reusing exploration information, such as simplex
tables and cutting planes, across the subproblems to significantly reduce overall
computational effort without compromising precision.
To validate the effectiveness of MOBB/D, we conducted a series of numerical
experiments that compared it with existing multi-objective branch-and-bound
methods. Our results demonstrate that MOBB/D not only achieves higher computational
efficiency but also ensures the quality of its solutions. Additionally, we applied
the proposed framework to a real-world case study involving waste collection in
the city of Muroran, where we aimed to balance operational efficiency (e.g., total
vehicle distance) and resident burden (e.g., walking distance to collection points).
In this practical setting, MOBB/D successfully identified high-quality Paretooptimal
solutions, outperforming conventional methods while remaining feasible for
policymakers. This highlights its potential for supporting evidence-based policy
making (EBPM), where multiple stakeholders need to collaboratively explore tradeoffs.
In conclusion, MOBB/D offers a new path toward reconciling the computational
costs and precision requirements inherent in MOMILP problems, representing a
substantial advancement in multi-objective optimization methodology. Future work
includes incorporating additional enhancements such as selective pruning of
exploration data, column generation techniques, and extending the scope to nonlinear
or more complex problems. Through these developments, MOBB/D can be expected
to further broaden its applicability, thereby contributing to more sophisticated
decision-making processes in both academic and practical realms. It stands to serve
as a valuable tool in a wide range of applications, from urban planning and supply
chain management to healthcare and environmental policy. MOBB/D underscores the
potential for innovative, decomposition-based exact methods to reshape the
landscape of multi-objective optimization.doctoral thesi
International Academic Exchange with University of Johannesburg from the Viewpoint of Research and Education regarding Family System and Human Rights
Get PDFThe purpose of this report is to summarize the current progress of the international joint research on the protection of minorities in Japan and South Africa. South Africa has enacted the world's longest Constitution (1996), which enshrines the rights of all people in the state and affirms the democratic values of human dignity, equality and freedom. The history of and situation after Apartheid, as well as the reconstruction of the social system to guarantee these values, are highly suggestive for Japanese policy regarding the protection of children, women, sexual minorities, and indigenous people (such as the Ainu people).departmental bulletin pape
北海道におけるニホンジカの自動個体識別と検出：農業保護のためのモジュール軽量化
Get PDF室蘭工業大学博士（工学）この研究では、ディープラーニングとカメラトラップ技術を用いて、北海道に生息するニホンジカに焦点を当て、野生生物のモニタリングと保護活動を強化する方法について取り組んだ。人の手による野生生物識別の課題に着目し、ディープラーニングモデルを用いて種を分類・識別する完全自動化システムを提案した。この研究では、まず、EfficientNetB0とVGG16という2つの畳み込みニューラルネットワーク（CNN）の複数種分類における性能を評価し、EfficientNetB0が精度、適合率、再現率においてVGG16を上回ることを発見した。このモデルは80%以上の精度を示し、トップ5の精度では90%以上を達成しており、野生生物モニタリング用途に非常に効果的であることを示した。
次なる研究として、野生生物モニタリングで多く利用される既製品の3つのカメラトラップの性能について評価し、ニホンジカの検出に最も効果的である製品がソーラーパワー4kトレイルカメラであることを明らかにした。この研究では、生態学研究におけるカメラトラップの性能を最適化するために、カメラのキャリブレーションと歪み補正が重要であることを強調しています。
さらに，本研究の重要なトピックとして、トリプレット損失関数を用いたサイアミーズネットワークアーキテクチャを用いた個々のエゾシカの識別にも焦点を当てました。複数のモデルを適用し比較した結果、ResNet152モデルは最も成功し、斑点パターンなどの特徴に基づいて個々のエゾシカを分類することで高い精度を達成しました。本研究ではまた、人間と野生生物の軋轢、特に北海道におけるエゾシカとクマによる農業被害についても調査した。これを緩和するために、Raspberry Piに展開されPIRセンサーを統合したYOLOv8-nanoモデルを使用して、エゾシカ検知システムを開発した。このシステムは、電子メールとLINEアプリを介して地域住民にリアルタイムの警告を提供し、音声通知も提供できる。ブザーによるシカの抑止効果はいま一つであったが、警告システムはシカの活動を監視することに成功し、保全戦略のための貴重な知見を提供した。将来の課題として、農業地域における野生生物管理の改善のために、ドローンなどの代替抑止策を探求するためのさらなる研究が推奨される。This study explores the use of deep learning and camera trap technologies to enhance wildlife monitoring and conservation efforts, with a focus on Sika deer in Hokkaido, Japan. It addresses the challenges of manual wildlife identification and proposes a fully automated system to classify and identify species using deep learning models. The study evaluates the performance of two Convolutional Neural Networks (CNNs), EfficientNetB0 and VGG16, for multi-species classification, finding that EfficientNetB0 outperformed VGG16 in terms of accuracy, precision, and recall. The model demonstrated over 80% accuracy and achieved a top-5 accuracy of over 90%, making it highly effective for wildlife monitoring applications. Additionally, the research assesses the performance of three camera traps, highlighting the Solar-Powered 4k-Trail camera as the most effective in detecting Sika deer. The study emphasizes the importance of camera calibration and distortion correction for optimizing camera trap performance in ecological research. A key part of the study also focused on the identification of individual Sika deer using a Siamese Network Architecture with triplet loss functions. The ResNet152 model was the most successful, achieving high accuracy in classifying individual Sika deer based on distinctive features like spot patterns. The study also investigates human-wildlife conflict, particularly the agricultural damage caused by Sika deer and bears in Hokkaido. To mitigate this, a Sika deer detection system was developed using the YOLOv8-nano model deployed on a Raspberry Pi, integrated with a PIR sensor. The system provided real-time alerts to local residents via email and the LINE application, alongside audio notifications. While the buzzer's effectiveness in deterring deer was mixed, the alert system proved successful in monitoring deer activities, offering valuable insights for conservation strategies. Further research is recommended to explore alternative deterrents, such as drones, for improving wildlife management in agricultural areas.doctoral thesi
生徒のメタ認知を促す学習環境：自己調整学習システムの電気科授業における活用事例
Get PDF室蘭工業大学Muroran Institute of Technology博士（工学）本論文は、教育研究と工学研究の複合領域的な取り組みである。特に人工知能（AI）と自己調整学習（SRL）に焦点を当てている。本研究では、まず、AIを教育研究に統合する際の課題について洗い出しを行い、従来の教育理論と技術進歩の間のギャップを浮き彫りにした。生徒の学習を促進するための様々なアプローチ、具体的にはゲームやデジタルツールを用いた生徒の行動と学習パターンの分析と検証を行った。ゲームをプレイする中での生徒たちの協働を観察した。ゲーム内の行動とアンケートの回答を比較することで、学習スタイルと問題解決傾向の分類を目指した。しかし、実際の教室での行動を捉えることの限界と、教科に基づく学習アプローチの多様性が課題として浮き彫りになった。また、学習の中で生徒自身が自身の特性などを意識して行動しているかというメタ認知についての重要性も認識した。
そこで，本研究は，自立学習スキルを育成するための枠組みとしてSRLへと方向転換した。具体的な学習環境として電気物理学の演習授業を対象とし、演習における実践的な活動と専門家の指導を用いて生徒の学習行動を評価した。機械学習技術を用いて学生のプロファイルを分類し、確立された教育心理学理論と整合する有望な結果を示した。
研究の最終段階では、デジタルツールからeラーニングプラットフォームへ移行することによってアプローチを洗練させ、よりスケーラブルで効率的なデータ収集を可能にした。機械学習によって分析された学習行動は、学習管理システム（LMS）やパーソナライズ学習アプリケーションに統合可能なSRLベースの学生プロファイルを生成するために活用される。本研究は、従来の心理学理論と現代のデジタル学習環境を結び付け、教育におけるAI活用のための構造化されたフレームワークを提示している。これは、AI主導の洞察を通じて、学生の学習成果を向上させることを目指す将来の研究の基盤となるとともに、教師、教育政策立案者、そしてテクノロジー開発者にとって実用的な示唆を提供します。This thesis explores the intersection of educational research and technology, particularly focusing on Artificial Intelligence (AI), metacognition, and Self-Regulated Learning (SRL). The study begins by addressing the challenges in integrating AI into educational research, highlighting the gap between traditional educational theories and technological advancements. The work investigates various approaches to enhance learning, including the use of games and digital tools to analyze student behavior and learning patterns. Initial efforts involved using questionnaires and game-based environments to explore learning styles and problem-solving tendencies. However, it became clear that questionnaire data did not reflect real classroom behavior, and games, though insightful, were impractical for curricular use. These issues led the research to embed behavior analysis directly into the learning process. Central to this transition was a deeper emphasis on metacognition—how learners monitor, reflect on, and regulate their thinking during tasks. Rather than relying on external tools detached from instruction, the study aimed to make metacognitive processes observable and instructive within authentic classroom settings. Metacognition and SRL were positioned not only as theoretical frameworks but also as practical guides for instructional design, enabling students to take ownership of their learning. To implement this, the research focused on an electrical physics course, using hands-on tasks and expert guidance to assess students' learning behavior in context. Machine learning techniques were applied to categorize student profiles, yielding results aligned with established educational psychology theories. In the final phase, the approach evolved by transitioning from isolated tools to an integrated e-learning platform, allowing for scalable and real-time data collection. The resulting learner profiles, grounded in SRL and metacognitive behavior, can be embedded into learning management systems (LMS) and adaptive learning applications. This research presents a structured framework for leveraging AI in education, linking psychological theory with digital environments. It provides a foundation for future studies aiming to improve learning outcomes through AI-driven insights, while offering practical implications for educators, policymakers, and developers focused on enhancing metacognitive growth and student agency.doctoral thesi
Highly confined out-of-plane corner states in air-hole type topological photonic crystal fiber
No full textOut-of-plane corner states are found in air-hole type square lattice topological
photonics crystal fiber (PCF) based on highly nonlinear glass. While, for the unit cell of
photonic crystal (PC) without lattice shift (UC1), the parities of Bloch modes under the first
photonic bandgap (PBG) are reversed between and X points in the Brillouin zone, the parities
are preserved for the unit cell with lattice shift (UC2). This is opposite compared with the case
of pillar type topological PCFs. Based on these results, out-of-plane optical states of the fiber
with three UC1 and one UC2 regions are investigated, and two highly confined corner states
are found at the corner of UC1 and UC2 interface. The corner states are not found by just
flipping unit cell, namely, for three UC2 and one UC1 PC corner (for pillar type PCF, the corner
states were found for this placement). Wavelength dependence of effective area and the
confinement loss of the corner states are investigated in detail, and by increasing the number of
layers, the confinement loss can be dramatically reduced. These new fibers may be applied to
nonlinear optical applications, such as wavelength conversion and single photon generation.journal articl
Comprehensive study of the luminescence properties of elemental metals
Get PDFThe lifetime and intensity of ultrafast luminescence in the near-infrared region were investigated for 15 representative elemental metals: simple light metals (Be, Mg, Al), 3d- (Ti, Ni, Zn), 4d- (Zr, Mo, Pd, Sn), 5d- (W, Pt) transition metals and noble metals (Au, Ag, Cu). The luminescence intensity at 0.9 eV was distributed over a range as wide as 2.5 orders of magnitude. The sum of an instantaneous response and an exponential decay component well approximated the time-evolution of the luminescence intensity at 0.6 eV. The lifetimes obtained from this decomposition were distributed in a narrower range from 110 fs (Be) to 694 fs (Ag), corresponding to a factor 6.3. We proposed a model based on the dielectric function of the bulk metal to understand the luminescence intensity of the quasi-instantaneous component. It was found that the luminescence intensity was inversely proportional to the Drude damping constant, that is, the slower the damping, the stronger the luminescence. This shows that quasi-instantaneous luminescence is quenched by the non-radiative decay of the non-thermal electrons via electron-electron scattering. The electron-phonon coupling strengths were evaluated using Eliashberg functions derived from ab initio calculations and the decay rates corresponding to electron-phonon scattering were evaluated using the extended two-temperature model considering the equilibration process of excited electrons. We found a fairly good agreement between the experimental and the theoretical results for the absolute values of the exponential decay rates.journal articl
Using PAC analysis to decipher an instructor's own impressions of the teacher license renewal courses
Get PDFFrom 2009 to 2022, a teacher license renewal system was implemented in Japan. Teachers were required to take the courses once every 10 years. Previous studies have revealed that teachers who took the courses positively evaluated the courses, but also expressed dissatisfaction with the heavy burden of taking the courses. However, there has been little discussion of the impressions of university teachers who lecture on the courses. In this paper, we used PAC analysis to listen to a university teacher’s impressions of the renewal courses and obtain advice from other university teachers to make the interpretation more multifaceted. University teachers made efforts to improve the renewal courses, but because participation in the courses were compulsory, it was difficult for their efforts to be rewarded. It has emerged that the new training system should ensure the autonomy of teachers and the autonomy of university teachers.departmental bulletin pape
Evaluating the Impact of Illustrated Activities on Student Engagement and Learning Outcomes
Get PDFThis article investigates the influence of a hand-drawn illustrated vocabulary activity on student engagement and learning outcomes in a university English Communication classroom. Conducted at Muroran Institute of Technology, the study involved 27 second-year undergraduate students participating in a modified Pictionary-style activity. The research aims to understand how drawn visual elements can enhance vocabulary retention and overall learning experiences from the student perspective. Survey responses, based on a Likert scale, revealed that students found these illustrated activities more effective for vocabulary retention compared to traditional methods, increased their confidence in using new words, encouraged active class participation, and made learning more enjoyable. These results suggest that hand-drawn illustrated vocabulary activities could be a valuable addition to English as a Foreign Language (EFL) instruction, promoting a more engaging, interactive, and supportive learning environment.departmental bulletin pape