Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
組織の意思決定のためのデータ共有の仕組みの構築
Get PDF九州工業大学では2003年度に紙ベースでの自己評価を開始し、2005年度に情報工学部でJABEE認定以降、自己評価を全学展開し、2007年度には学修自己評価システムの開発を始めた。2009年度には正課外活動も記録できるeポートフォリオを全学的に稼働させ、現在に至るまでシステムの改良を続けながら全学に定着させてきた。このように教学データが集められ個人にフィードバックする仕組みが存在している一方で、本学では組織として意思決定をするためのサポートの仕組みは存在しておらず、2022年度より教学IR（Institutional Research）の仕組みを強化すべく本学の学習教育センターに教学IR支援グループを置くこととなった。
学習自己評価システムだけでなく、学内にはさまざまなデータが存在しているが、これらのデータを活用するための統合的なシステムは存在せず、各担当部署によって管理している。そこで、本学ではまずIRに関する最初のシステムとしてデータを結合して活用するためにETL（Extract Transform Load）ツールを導入した。続いて、集中レポジトリとしてデータレイクシステムを導入した。
本稿ではIRの概要、ETLツールの導入やデータレイクの構築など、本学で推進しているIRシステムについて紹介する。最後に、データウェアハウスやビジネスインテリジェンスツールの導入も視野に入れた今後の展開について述べる。departmental bulletin pape
小径穴加工プロセス状態の深層特徴の多変量正規分布に基づく指標化と可視化および工具状態認識への応用
Get PDF九州工業大学博士（情報工学）第1章 緒論| 第2章 穴加工の監視技術| 第3章 深層学習モデル| 第4章 実験方法| 第5章 解析手法| 第6章 小径穴加工の状態推移監視| 第7章 結論工具摩耗は，寸法精度や表面品質の低下を引き起こし，同時に予期せぬ工具破損の原因となる．穴加工中に工具が破損すると，高価なワークピースに修復不可能な損傷を与え，操業コストの増加や生産遅延に繋がる．特に小径工具は負荷に対して機械的強度が不足しており折損しやすいため，安定した加工を継続することが難しい．生産性を向上させ，加工工程の自動化や工作機械の知能化を推進するためも，工具状態の診断と監視が非常に重要である．
本論文では，時系列信号解析における画像レベルの異常検出と定量的指標化手法であるDDM（Deep feature Distribution Modeling）を提案した．穴加工中に計測したAE（Acoustic Emission）信号を，連続ウェーブレット変換で前処理し，事前学習済みの深層モデルを用いて特徴量を抽出した．得られた深層特徴に多変量正規分布を当てはめることで正規性を確立し，異常スコアとして定量的に指標化した．さらに，提案手法では，複数の深層モデル層から得られた異常マップを生成することにより，より細分化されたレイヤーコンテキストを用いて異常状態の違いを可視化することが可能となる．小径穴加工中のAE信号計測を自動化する計測システムの開発と測定・評価実験による検討を行い，本測定手法の有効性を実証した．
第1章『緒論』では，本研究の背景と従来の工具状態監視（Tool Condition Monitoring ; TCM）における課題を概観し，提案手法の特徴と効果について述べ，本研究の立場を明らかにした．
第2章『穴加工の監視技術』では，ドリルによる穴加工の基礎とTCMに用いられる幾つかのセンシング技術について述べ，穴加工中に計測されるAE信号特性を示した．
第3章『深層学習モデル』では，提案手法の基礎となる深層学習モデルの構造とその学習方法を概観し，実運用におけるモデル学習の課題やより効果的な状態監視への適用方法について検討した．
第4章『計測システムの開発と実験方法』では，小径穴加工試験の実験方法とAE信号計測を自動化する計測制御システムの構成を紹介した．また，AE信号の伝搬特性に関する検証試験を行い，加工位置などがAE信号強度に与える影響を調査した．
第5章『解析手法』では，深層特徴のデータ分布に基づき異常状態を定量化する提案手法について解説し，ドリルの損傷に対する提案手法の識別性能を検証した．
第6章『小径穴加工の状態推移監視』では，工具折損に至るまでの，工具摩耗推移と切りくずの形態の変遷と観察し，評価指標との対応関係を確認した．さらに，解釈性を向上させるために状態指標の可視化を試み，状態監視における提案手法の有効性が示唆された．
第7章『結論』では，各章で得られた結論をまとめ，本論文の成果を総括した．九州工業大学博士学位論文 学位記番号: 情工博甲第393号 学位授与年月日:令和6年3月25日令和5年度doctoral thesi
Flexibility and Productivity in IoT Programming: A Case Study with Mruby
Get PDFThis paper explores the use of data flow-based design in IoT application development and its integration with the Node-RED programming tool. Node-RED provides a visual interface for creating programs based on the flow of data between nodes. However, Node-RED is not suitable for resource-constrained devices like micro-controllers. To address this, we propose a method of generating mruby code from Node-RED’s data flow programs. The generated mruby code can be compiled into device-independent byte-code and executed on the mruby VM. To facilitate the implementation, we introduce the RBoard, a prototyping micro-controller board equipped with the mruby VM firmware. This approach combines the intuitive data flow design of Node-RED with the power of mruby for embedded systems. Through this integration, developers can rapidly prototype IoT applications and leverage the flexibility of scripting languages. Our findings demonstrate the feasibility and benefits of using data flow-based design and mruby code generation in IoT application development. This research contributes to the advancement of efficient and scalable IoT programming methodologies and enables the creation of innovative IoT solutions on micro-controller platforms.International Conference on Advanced Research in Technologies, Information, Innovation and Sustainability, ARTIIS 2023, October 18–20, 2023, Madrid, Spainjournal articl
オープンソースのリモートセンシングデータと機械学習アプローチを活用したラオス人民民主共和国ナム・グム川流域（NNRB）における洪水危険度マッピング
Get PDF九州工業大学九州工業大学博士学位論文（要旨）学位記番号: 工博甲第594号 学位授与年月日:令和6年9月25日thesi
テストによる歩留まり低下を軽減するためのIRドロップ考慮型テストに関する研究
Get PDF九州工業大学九州工業大学博士学位論文（要旨）学位記番号: 情工博甲第405号 学位授与年月日:令和6年9月25日thesi
効率的な大量生産のための1Uモジュール式キューブサットインターフェースの開発と標準化
Get PDF九州工業大学博士（工学）1 Introduction| 2 CubeSat and standard interfaces| 3 Tools for evaluating Product design performance| 4 Design and Development, Evaluation| 5 Standardizing CubeSat architecture| 6 Verification and validation: Analysis and Testing campaign| 7 Results and Discussion| 8 Conclusion| 9 ReferenceAdvancements in CubeSat technologies are transforming both space research and commercial satellite applications. These innovations have greatly improved CubeSat missions' capabilities, with modern CubeSats now featuring advanced instrumentation, enhanced communication systems, and improved propulsion technologies. Consequently, CubeSats can undertake more complex missions that were once the domain of larger satellites, including Earth observation, astronomical research, and interplanetary missions. A notable trend is the increasing interest in deploying constellations of CubeSats for comprehensive coverage and continuous monitoring. The commercial viability of CubeSats in constellation missions is also rising as their capabilities expand. Private space firms are becoming key players in this field, attracted by the relatively low cost of deployment, the potential for rapid technology updates, and scalable constellations. As demand for CubeSat constellations grows, so does the need for their mass production.
Efficient and standardized design and development processes are crucial for producing CubeSats in large quantities at a reduced cost. The efficiency of these processes is vital to meeting time and budget constraints. Any inefficiency in the product development process can amplify the effects of these constraints, especially when producing items in large quantities. A primary source of inefficiency is the complexity of satellite configuration during design, testing, and upgrading, which impacts project time and cost.
Several factors contribute to the complexity of CubeSat assembly and deployment. The traditional CubeSat mechanical interface uses a stacking arrangement, where systems are layered, leading to interconnected internal subsystems. This increases interdependency during assembly, requires numerous structural parts, and raises both assembly time and production costs. Unlike the standardized external interface with the launch vehicle, CubeSat internal interfaces lack standardization.
This lack of standardization presents several challenges. Commercially available CubeSat subsystems often have unique interface requirements, making them incompatible with other subsystems. Customizing these interfaces to meet specific mission requirements significantly extends development time and costs due to the need for design validation and testing. Testing and upgrading CubeSats with non-standard interfaces is also more complex and time-consuming, as each satellite requires unique configurations, complicated testing and repairs. Incompatibility issues between subsystem interfaces can sometimes force designers to compromise on mission objectives. Custom designs generally exhibit lower reliability, with a higher likelihood of errors and failures during integration. Mechanical and electrical mismatches can cause critical mission issues, further complicating the design and integration process. Non-standardized interfaces also limit flexibility, making it difficult to swap out or upgrade subsystems easily. This lack of flexibility hinders the ability to adapt CubeSats for different missions and quickly incorporate new technologies.
These challenges highlight the importance of developing standardized and versatile interfaces to facilitate easier integration and enhance mission success. To address these inefficiencies, it is crucial to evaluate design complexity and efficiency for higher performance in mass production. Design performance evaluation tools used in industrial products, such as Design for Manufacturing and Assembly (DFMA) and complexity analysis, can provide valuable insights during the initial development phase. Implementing these tools in satellite systems can mitigate design and manufacturing-related issues, enhance assembly performance, and reduce costs.
Design modularity plays a significant role in providing flexibility among subsystems for easy assembly, testing, and upgrades. This study aims to investigate the efficiency of various CubeSat platforms' mechanical interfaces by applying industrial design evaluation tools and developing a modularized, simple, and flexible platform. The goal is to standardize the developed platform by defining an interface that maximizes the efficiency of existing interfaces. This standard interface aims to address compatibility issues among different vendors, enhance design flexibility, and ultimately lead to more efficient and successful missions, particularly in cost- and time-sensitive mass production CubeSat missions.九州工業大学博士学位論文 学位記番号:工博甲第592号 学位授与年月日:令和6年9月25日令和6年度doctoral thesi
Individual adaptation and social attributes in a handshake robot with CPG control
Get PDFIn a diverse society, a crucial aspect of developing social robots is their ability to adapt to each individual's characteristics. There are two types of adaptation: physical and social. In pHRI, attention is paid to physical adaptation, but by investigating the social meaning of physical adaptation, we can extend it to social adaptation. Although it is a physical interaction, a handshake is a social action with complex meanings. We realized a handshaking robot with a Rowat-Selverston CPG controller that can synchronize with human movements. This study aims to clarify the relationship between physical and social adaptation by examining individual characteristics inferred from the internal state of the robot and human-robotic social attributes. Our finding is that there is a correlation between the internal state of the robot and some robotic social attributes. It was also found that the length of the handshake had little effect on these social perceptions. This study highlights the complex interplay between robot physical adaptability and social attribution, providing a foundation for developing robots capable of personalized and socially meaningful interactions.journal articl
A network-based trans-omics approach for predicting synergistic drug combinations
Get PDFBackground
Combination therapy can offer greater efficacy on medical treatments. However, the discovery of synergistic drug combinations is challenging. We propose a novel computational method, SyndrumNET, to predict synergistic drug combinations by network propagation with trans-omics analyses.
Methods
The prediction is based on the topological relationship, network-based proximity, and transcriptional correlation between diseases and drugs. SyndrumNET was applied to analyzing six diseases including asthma, diabetes, hypertension, colorectal cancer, acute myeloid leukemia (AML), and chronic myeloid leukemia (CML).
Results
Here we show that SyndrumNET outperforms the previous methods in terms of high accuracy. We perform in vitro cell survival assays to validate our prediction for CML. Of the top 17 predicted drug pairs, 14 drug pairs successfully exhibits synergistic anticancer effects. Our mode-of-action analysis also reveals that the drug synergy of the top predicted combination of capsaicin and mitoxantrone is due to the complementary regulation of 12 pathways, including the Rap1 signaling pathway.
Conclusions
The proposed method is expected to be useful for discovering synergistic drug combinations for various complex diseases.journal articl
PredIL13: stacking a variety of machine and deep learning methods with ESM-2 language model for identifying IL13-inducing peptides
Get PDFInterleukin (IL)-13 has emerged as one of the recently identified cytokine. Since IL-13 causes the severity of COVID-19 and alters crucial biological processes, it is urgent to explore novel molecules or peptides capable of including IL-13. Computational prediction has received attention as a complementary method to in-vivo and in-vitro experimental identification of IL-13 inducing peptides, because experimental identification is time-consuming, laborious, and expensive. A few computational tools have been presented, including the IL13Pred and iIL13Pred. To increase prediction capability, we have developed PredIL13, a cutting-edge ensemble learning method with the latest ESM-2 protein language model. This method stacked the probability scores outputted by 168 single-feature machine/deep learning models, and then trained a logistic regression-based meta-classifier with the stacked probability score vectors. The key technology was to implement ESM-2 and to select the optimal single-feature models according to their absolute weight coefficient for logistic regression (AWCLR), an indicator of the importance of each single-feature model. Especially, the sequential deletion of single-feature models based on the iterative AWCLR ranking (SDIWC) method constructed the meta-classifier consisting of the top 16 single-feature models, named PredIL13, while considering the model’s accuracy. The PredIL13 greatly outperformed the-state-of-the-art predictors, thus is an invaluable tool for accelerating the detection of IL13-inducing peptide within the human genome.journal articl
Experimental Exploration of the Power of Conditional GAN in Image Reconstruction-Based Adversarial Attack Defense Strategies
Get PDFAdversarial attacks pose a significant threat to the reliability and security of deep learning models, particularly in image processing applications. Defending against these sophisticated manipulations requires innovative strategies, with Generative Adversarial Networks (GANs) emerging as a promising solution. This paper presents an experimental exploration of the power of conditional Generative Adversarial Networks (cGANs) in image reconstruction-based strategies for defending against adversarial attacks. Our study involves a comparative analysis of four distinct image reconstruction models: the traditional GAN-based Defense-GAN, the cGAN-based method exemplified by pix2pix, a hybrid approach combining pix2pix with perceptual loss, and a generator model centered around residual blocks. The results of our experiments demonstrate that cGAN models exhibit significantly enhanced efficacy in defending against adversarial attacks compared to other image reconstruction methods. This superiority is attributed to the inherent characteristics of cGANs, which we delve into in detail. The findings provide crucial insights for developing more robust defense strategies against adversarial attacks in diverse image processing and machine learning applications.journal articl