Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
Anomaly Detection from IGBT Images Based on VGGNet
Get PDFPower devices are semiconductor devices used for power control. They handle high voltages and large currents and are used in electric vehicles, televisions, and trains. These devices are also used in rice cookers, microwave ovens, air conditioners, etc. In homes, they are used in rice cookers, microwave ovens, air conditioners, and so on. Since power devices are electronic components closely connected to our daily lives, they must be highly reliable and safe. To ensure this, power cycle tests are conducted. During these tests, devices are subjected to electrical and thermal stress by repeatedly switching the power on and off. This simulates actual operation and analyzes the breakdown process of joints and other parts on the device chip. However, conventional tests generate sparks during the breakdown process. This severely damages the diode, which is the main part of the power device. This makes it difficult to identify the cause of the breakdown and analyze the process leading up to it. To solve problems in conventional power cycle testing, a new technology using ultrasonic observation is being developed. This technology makes it possible to output images of the test object in real time. Consequently, it is possible to continuously record structural changes inside the device during testing, which enables identification of failure causes and process analysis. There have been problems with conventional testing. However, there are still some issues to be resolved before the new technology can be used in practice. The main issues are the lack of an established method for analyzing large amounts of image data and extracting small changes in image features that are difficult to discern with the naked eye. In this paper, we propose a method for classifying ultrasound images obtained from tests using deep learning. Our method uses the pre-trained VGG16 model and introduces a new network model with an additional skip connection. This allows for the detection of subtle changes in images. Furthermore, we expand the dataset using CycleGAN and Mixup. These techniques reduce the influence of data bias and enable accurate image classification. In our experiments, we applied the proposed method to 201 ultrasound images and achieved a discrimination performance of precision = 0.9767, recall = 0.8936, and F-measure = 0.9333.journal articl
Pattern Formation of Mobile Agents in Dynamic Grids
No full textIn this paper, we consider the pattern formation problem of mobile agents in n× m dynamic grids, requiring k agents in the grid to stay at x designated (target) nodes to form some shape (pattern). We assume that at most one link is missing at each round. In this case, some agent’s movement may be always blocked and the agent cannot reach any target node. In addition, if the number k of agents is smaller than the number x of target nodes, several target nodes cannot be occupied. For this reason, focusing on the relationship between the values of k and x, we consider variants of the pattern formation problem in dynamic grids and examine how differences in these requirements and the number of agents influence the design and performance of algorithms. First, we consider the case of k ≤ x. In this case, we consider the approximate pattern formation problem, requiring at least k-1 (＜ x) target nodes to be occupied. For this problem, we propose an algorithm to achieve approximate pattern formation in O(kn + km) rounds. Next, we consider the case of k ＞x. In this case, we consider the exact pattern formation problem, requiring each of the x target nodes to be occupied. For this problem, we propose an algorithm to achieve exact pattern formation in O(kn + km + mn) rounds. In particular, when k – x is sufficiently large, we show that exact pattern formation can be achieved in Θ(n + m) rounds.journal articl
Two-Stage Reservoir Computing for Sensor-Specific Activity Recognition Using the WEAR Inertial Dataset
No full textapproach to address robustness and generalization in the 2nd WEAR Dataset Challenge of UbiComp HASCA 2025. The main task involves recognising 18 activities from inertial data collected at four locations in the body. The challenge emphasises generalisation and robustness to randomly sampled, sensor-specific 1-second windows from unseen participants, with augmentations simulating diverse wearing conditions. Prior baseline approaches, such as DeepCon- vLSTM and TinyHAR, struggle with low-motion or overlapping classes with limitations on handling short windows, and noisy inputs. In response, we propose a two-stage hybrid pipeline com- bining CNNs for spatial feature learning and Reservoir Computing (RC) for short-term temporal dynamics. To evaluate robustness and generalization, we compared our RC-based approach with three gradient boosting models and evaluated four RC variations. Experi- mental results show that the merged-limb two-stage RC strategy achieves the highest test macro F1 score (0.52888), outperforming the three challenge baselines and tree-based models. This study demonstrates the potential of reservoir computing across different granularities and architectures for building robust HAR models.journal articl
DNAナノテクノロジーと分子ロボティクス
No full text二重らせんという明確な構造と配列の相補性による二本鎖形成という大きな特徴をもつDNAは、ナノスケールの正確な構造体を作成する上で優れた素材となっている。本稿では、今日DNAナノテクノロジーにおける主要な技術となったDNAオリガミ法と、これを活用して構築できる分子マシン、さらにはDNAオリガミの医療分野への応用例について紹介した後に、これらを分子ロボットとつなぐDNAコンピューティングについても解説する。journal articl
Cell surface display of a protein based on a tail-anchored membrane protein
No full textMethods for displaying proteins on the cell surface are widely used in protein-based biotechnology and bioengineering, where target proteins are expressed as fusion constructs with membrane proteins through recombinant DNA technology. In this study, we developed a system for displaying a protein on the cell surface using the transmembrane domain (TMD) of a tail-anchored membrane protein (TA protein). TA proteins have an orientation in the cell membrane such that their C-termini are displayed on the cell surface, which contrasts with that of type I transmembrane proteins that are commonly used as anchoring units. Therefore, by utilizing the TMD of a TA protein as an anchoring unit, desired proteins can be attached to the TMD via their N-termini. This approach is advantageous for displaying proteins whose C-terminal regions play important roles in their activity. In this study, we chose the inner nuclear membrane protein emerin as a TA protein and constructed expression systems in mammalian cells for a series of fusion proteins based on deleted forms of emerin. We found that utilizing emerin that lacks 210 residues from the N-terminus as a TMD allowed efficient translocation of the fusion protein to the plasma membrane, successfully displaying its target protein portion on the cell surface. Thus, our system serves as an effective method for protein display, enhancing the applicability of cell surface display technology based on transmembrane proteins.journal articl
Comparative Analysis of Grade Distributions in Team-Taught Introductory Data Science Courses for First-Year Students
No full textThe increasing importance of mathematical and data science education in Japanese higher education has led to its integration into general education curricula. This study examines the design and implementation of a university-wide information literacy education course at Hokuriku University, focusing on the integration of data science and AI education. As part of the university’s general education curriculum, the compulsory first-year course aims to equip students with essential information literacy and data science skills. This study analyzes the fairness of evaluation among course instructors, considering the challenges of delivering uniform content to several students. These findings highlight the importance of regular communication and collaboration among instructors to ensure consistent grading practices. The innovative approach of the course to integrate data science and AI education into information literacy education serves as a model for other higher education institutions in Japan. By sharing the lessons learned from this experience, this study contributes to the development of best practices for designing and delivering university-wide general education courses that effectively prepare students for an increasingly data-driven world.journal articl
Improvising the Tesla Micro-Turbine: Experimental and Performance Analysis
Get PDFPresent Tesla turbine presented a distinct approach to conventional turbine technology by utilizing frictional and viscous forces acting on stacked discs, deviating from the traditional use of blades. Characterized by its efficient energy extraction through boundary layer adhesion and cohesion, the Tesla turbine can achieve exceptional rotational speeds. However, its limited torque output remains a significant challenge, particularly for applications demanding high torque. Hence, the objective of this manuscript is to improvise the design of the Tesla turbine and analyze the impact of the new design’s rotational speed, voltage, current and electrical power generation with increasing inlet pressure. The research methodology employed in this study involves the design of the Tesla turbine discs, then proceeded with experimental work from fabrication of the designed discs and conducting performance tests. It aims to gather quantitative data through controlled experiments to assess the best design of the turbine and test the performance and feasibility of the improvised Tesla turbine. The model comprises of bladed design around the disc performed the best at producing electrical power up to 19.08 W at 50 psi inlet pressure.journal articl
濃度マランゴニ効果を考慮したサイトカイン濃度勾配下での好中球の水中推進機構の研究
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: 生工博甲第504号 学位授与年月日:令和7年3月25日thesi
