Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
Logistic-Function Shaping of Radial Force for Acoustic Noise Reduction in Switched Reluctance Motor
Full text linkIn this paper, radial-force shaping using the logistic function is proposed for acoustic noise reduction in switched reluctance motors. The proposed radial-force shaping regulates the radial force on each stator tooth to track a logistic function with respect to rotor position. The logistic function is characterized by rising from zero to a constant value. This feature is used to define the radial-force references for the three phases so that their summation is always flat. By flattening the radial-force sum, the method effectively eliminates the multiples of the third component of the stator-tooth radial force. These components excite stator vibration in the form of mode 0. Consecutively, the proposed method significantly reduces the vibration and acoustic noise associated with mode 0. Furthermore, torque ripple is minimized through numerical sweeping involving only two parameters within a restricted search space. Experimental validation and finite element analysis confirm the effectiveness of the proposed method.journal articl
Kinetic Modeling of Ammonia and Hydrogen Dissociative Co-adsorption on Iron Surface and its Effect on Hydrogen Embrittlement
Full text linkWe study the mitigation of environmental hydrogen embrittlement of iron by ammonia impurities in the hydrogen gas using combined theoretical and experimental methods. The competitive and dissociative co-adsorption of gaseous ammonia and gaseous hydrogen on the iron surface was investigated using density functional theory. The surface adsorption and decomposition of ammonia, as well as the ammonia partial pressure, were considered as influential factors contributing to the control of atomistic hydrogen uptake into the material. To elucidate the mechanism of ammonia competing with hydrogen on the iron surface and of ammonia mitigating hydrogen embrittlement, we develop kinetic modeling that can estimate the reaction rate and the dynamic surface coverage of different adsorbed species on the iron surface. The reaction rates for hydrogen and ammonia co-adsorption and dissociation were calculated using transition state theory combined with the Langmuir adsorption model, and a fracture toughness test was conducted to validate theoretical results. The adsorption rate of ammonia on iron is significantly higher compared to hydrogen, thus, ammonia hinders the hydrogen adsorption on the iron surface. However, partial pressure dependent ammonia decomposition also provides hydrogen atoms, which induce hydrogen embrittlement. The theoretical results of this study were supported well by experimental fracture toughness test results.journal articl
Numerical Simulation of Output Filter Design for Optoelectronic Deep Neural Network Using Hyperparameter Optimization
Full text linkconference pape
Effects of severe shear strain and hydrostatic compression on the crystal structure and magnetism and of manganese oxide Mn3O4
Full text linkThe effects of severe shear strain and hydrostatic compression on the magnetostructural correlation of manganese oxide Mn3O4 were investigated using magnetic measurements and x-ray diffraction analyses. Hydrostatic pressures up to PHP = 4 GPa were applied using a diamond anvil cell, whereas shear stress was applied through high-pressure torsion (HPT) processing at PHPT = 6 GPa. A correlation between the coercive field and Jahn–Teller distortion was observed. Hysteresis measurements for magnetization showed a sharp reduction in the coercive field starting from PHP = 2 GPa under hydrostatic pressure. By contrast, an increase of 40% was observed after one rotation (N = 1) during HPT processing. The Jahn–Teller effect was evaluated with the ratio of the long to short distances of the MnO6 octahedron, which remained constant at 1.24 up to PHP = 2 GPa and then increased to 1.26 at PHP = 3 GPa. However, under HPT processing, this ratio changed from 1.24 to 1.27 at N = 1 and decreased to 1.18 at N = 3. The Mn2+–O–Mn3+ bonding angle also changed according to the Jahn–Teller distortion in both compression modes. Thus, the local symmetry of the MnO6 octahedron and MnO4 tetrahedron was modified through compression and subsequent shear stress, allowing for control over the hysteresis of magnetization.journal articl
Functional role and folding properties of the glucan-binding domain of oral bacterial glucansucrase
Full text linkA Streptococcus sobrinus glucansucrase (i.e., glucosyltransferase-I) elongates α-1, 3-linked glucose chains at pre-existing branch points of dextran, and its activity is enhanced by primer dextran. The glucansucrase contains a catalytic domain and a glucan-binding domain (GBd) with six tandem repeats. We here examined the role of GBd in the glucansucrase, focusing on its impact on enzymatic activity, dextran binding, and structural stability. By generating seven deletion mutants and a circularly permuted protein, our research demonstrates that the first four tandem repeats in the GBd and proper domain orientation are required for efficient glucosyl transfer. Moreover, characterization of circular dichroism and thermal unfolding indicated that glucosyl transfer efficiency is linked to cooperative interdomain folding. These findings highlight the importance of interdomain allostery in optimizing glucansucrase function.journal articl
AURA-MFM: Activity Understanding and Representations Alignment - Multimodal Foundation Model
Full text link近年，ウェアラブルデバイスの普及に伴い，IMUを用いた行動解析が重要性を増している．医療やロボティクスなど広範な応用が期待される中，最近の研究ではユニモーダル解析だけでなくマルチモーダル解析への関心が高まっている．いくつかの既存研究では一人称視点動画やテキストなどを導入したマルチモーダル基盤モデルが提案されているが．人間の全身動作の詳細な解析には至っていない．そこで，本研究では新たに三人称視点動画とモーションキャプチャを導入し，IMUとテキストを加えた4つのモダリティから成る基盤モデルとして，AURA-MFM(Activity Understanding and Representations Alignment - Multimodal Foundation Model)を提案する．三人称視点動画とモーションキャプチャデータを用いることで，一人称視点では捉えきれない人間の全身動作の詳細かつ多角的な解析を行うことを目指す．また，TransformerベースのIMUエンコーダを採用することで，IMUデータの複雑な動作情報を精緻に学習し，性能向上を図る．実験ではリトリーバルタスクと行動認識タスクにより評価を行い，既存手法を上回る性能が得られた．特に，行動認識タスクのゼロショット分類では，既存手法の結果(F値: 0.0747, Accuracy: 0.1961)と比較して大幅に上回る分類性能(F値: 0.6226, Accuracy: 0.7320)が得られた．In recent years, the widespread adoption of wearable devices has highlighted the growing importance of behavior analysis using IMU. While applications span diverse fields such as healthcare and robotics, recent studies have increasingly focused on multimodal analysis, in addition to unimodal analysis. Some existing research has proposed multimodal foundational models incorporating first-person video and text data; however, these models have yet to achieve detailed analysis of full-body human activity. To address this limitation, we propose AURA-MFM, a foundational model integrating four modalities: third-person video, motion capture, IMU, and text. By utilizing third-person video and motion capture data, the model enables a detailed and multidimensional understanding of human activity, which is difficult to capture with first-person perspectives alone. Additionally, a Transformer-based IMU encoder is employed to enhance the model's overall performance. Experimental evaluations conducted on retrieval tasks and activity recognition tasks demonstrate that our proposed model outperforms the existing method. Notably, in the zero-shot classification for action recognition, our method achieved significantly higher performance (F1-score: 0.6226, Accuracy: 0.7320) compared to the existing method (F1-score: 0.0747, Accuracy: 0.1961).journal articl
7-Methylquinolinium Iodobismuthate Memristor: Exploring Plasticity and Memristive Properties for Digit Classification in Physical Reservoir Computing
Full text linkThis study explores 7-methylquinolinium halobismuthates (I, Br, and Cl) with a focus on: (1) the impact of halide composition on their structural and semiconducting properties, and (2) their memristive behavior and plasticity for neuromorphic and reservoir computing. Crystallographic analysis reveals that halide substitution leads to the formation of low-dimensional bismuth-based frameworks. Optical bandgaps, measured via diffuse reflectance spectroscopy, correlate well with density functional theory calculations. Due to solubility constraints, only bismuth iodide complexes are integrated into electronic devices. Current–voltage measurements reveal pinched hysteresis loops, indicative of memristive behavior. Conductivity versus temperature analysis suggests both ionic and electronic conduction pathways. Given their ability to function as synaptic analogs, further tests on transient conductance (potentiation–depression) and spike-time-dependent plasticity are performed. The observed nonlinearity and fading memory characteristics highlight their potential for physical reservoir computing. To evaluate system-level behavior, a device with 16 gold electrodes (1 input, 15 outputs) is fabricated on a silicon substrate coated with the target compound. The device is assessed through benchmark tasks including waveform generation, NARMA-2, memory capacity, and noise response under DC/AC signals. Finally, the system demonstrates 82.26% accuracy in MNIST digit classification and 82% accuracy in spoken digit “2” recognition across six different speakers.journal articl
Data assimilation between Monte Carlo simulation and experimental results of crystal growth island density in REBCO thin films during the initial growth stages
Full text linkYBCO superconductors offer high transition temperatures for applications. REBCO wires, made via pulsed laser deposition (PLD), are developed for various uses. However, optimizing PLD parameters is time-consuming. To resolve this, we aimed to develop the simulation of REBCO thin film growth with Monte Carlo algorithm and Bayesian optimization to find the optimal parameters. While current simulations show discrepancies with experiments, this approach sets the foundation for improved models and simulations.journal articl
Magnetic Properties, Heat Generation, and Apatite Formation Ability of Mg-Ti Ferrite Particles Synthesized by Solid-State Reaction and Polymerized Complex Methods
Full text linkTitanium-doped ferrite has garnered significant interest as thermoseeds for cancer hyperthermia because of its controllable Curie point near body temperature, which prevents overheating and ensures high biological safety. However, few studies examine the effect of the synthesis conditions on microstructure, magnetic properties, and heat generation in an alternating magnetic field. Herein, Mg1+xFe2−2xTixO4 (x = 0.35, 0.45) particles are synthesized by a solid-state reaction and polymerized complex methods, followed by sintering at various temperatures. Their magnetic properties and heat generation behavior in an alternating magnetic field are investigated. Particles with x = 0.45 generate significantly less heat than those with x = 0.35, despite both being single-phase ferrite. Particles synthesized by the polymerized complex method at a sintering temperature of 1200 °C exhibit lower saturation magnetization but higher temperature increases compared with the solid-state reaction method. Additionally, in the sintering temperature range of 800–1000 °C, a temperature increase of more than 10 °C is observed in the polymerized complex method, likely a result of the inclusion of highly crystalline superparamagnetic particles. Furthermore, the ferrite particles form bone-like apatite on their surface in simulated body fluid, suggesting their potential as a novel material combining hyperthermia and bone integration properties.journal articl