Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
Classification of Histological Types of Primary Lung Cancer from CT Images Using Clinical Information
Get PDFIdentification of primary lung cancer is very important because it influences the course of treatment, especially for small cell carcinomas, which metastasize rapidly and have to be detected at an early stage. In addition to imaging, clinical information is often used in CAD (computer aided diagnosis) systems. Especially, clinical information such as smoking history, which is considered to be important in the diagnosis of lung cancer. In this paper, we propose a method to identify primary lung cancer by adding clinical information from medical records in addition to images to improve the accuracy of diagnosis. We use tumor images surrounded by rectangular regions from CT images in an open dataset as input images and train the method based on deep learning technique. We evaluate the proposed method by discriminating tumors from unknown data. In our experiments, we found that the accuracy was improved about 6% when clinical information was added to 604 images, which included four classes of cancer; adenocarcinoma, small cell carcinoma, squamous cell carcinoma, and large cell carcinoma.journal articl
Switching Waveform Design with Gate Charge Control for Power MOSFETs
Get PDF九州工業大学博士（工学）1 Motivation for switching waveform design| 2 Switching trajectory from gate charge perspective| 3 Gate charge controller| 4 Small gate charge pulse response| 5 Experimental demonstration of switching waveform design| 6 Discussion and future works| 7 ConclusionPower electronics control and convert electricity, and are currently attracting considerable attention as a means of saving energy owing to increasing public concerns regarding climate change. Driving power devices with optimal switching waveforms in a power converter realizes stabilized electric machine drive systems operation, downsizing, and high power density.
Active gate drives improve the switching behavior of power devices by incorporating active circuits into a conventional gate driving circuit. The use of digital hardware in the active gate drives allows precise waveform tuning and flexible operation, effectively accommodating individual variations in the devices or circuits. While the technology facilitates the generation of optimal switching waveforms due to the exponential increase in the number of possible candidates of gate drive settings, searching for the optimal waveform requires significant computational resources and time corresponding to the number of candidates. A new switching waveform design method is required to improve the efficiency of searching.
While the shape of the switching waveform that ensures system stability can be determined analytically, the iterations of the following processes have been employed to identify the optimal waveform: a power device is driven with a specified gate drive waveform, physical quantity is observed during its switching, the quantity is compared with the target value, and the device is driven with another gate drive waveform when a finite discrepancy exists between the measured and the target values. Machine-based auto-searching algorithms, which are currently attracting considerable attention nowadays, can mitigate the workload to some extent to determine the optimum. However, their stochastic nature results in the execution of unnecessary trials, which prevents the reduction of the searching workload. Developing a different approach from the flow described above is required to improve the searching efficiency.
This study presents a novel method for the design of switching waveforms. It uses a deterministic approach to resolve the gate drive waveform from the switching waveform. This study focuses on developing a method to determine the gate charge change required to attain the designed switching waveform from the observed waveform, considering that the gate charge affects the operation of power devices and that digital hardware provides arithmetic and memory functions. Consequently, it becomes crucial to identify a technique to establish the correlation between the switching voltage or current and the gate charge during the switching event of a power device.
The drain-source voltage and drain current during the switching of a power metal oxide semiconductor field effect transistor (MOSFET) are expressed as functions of gate charge in accordance with Kirchhoff’s law and semiconductor device physics. The gate charge waveform can be solved analytically for the designed switching waveform when these functions are explicitly defined, and the change in gate charge is not considerable. In practice, the relationships between the drain-source voltage or the drain current and the gate charge are generally unknown due to the variations incurred in individual devices. Nevertheless, the relationship can be provided discretely by examining the voltage and current response to the gate charge based on the following two assumptions. Firstly, the drain-source voltage and the drain current are assumed to respond instantaneously to changes in the gate charge. Secondly, the drain-source voltage, drain current, and gate charge can be approximated by piece-wise linear functions represented by linear combinations of linearly independent basis functions.
In this study, a set of triangular pulse shape functions is adopted as the basis for the gate charge to minimize and localize its influence on the switching waveforms of the drainsource voltage and the drain current. These triangular gate charge pulses are applied with varying time shifts during the switching event to observe how the drain-source voltage and drain current respond. The gate drive waveform required to achieve the desired switching behavior is then calculated based on the observed responses and the difference between the designed and actual switching waveforms. The validity of the proposed method is verified first with the simulation aided by SPICE. The simulated waveform of the MOSFET driven by the gate drive waveform computed using the proposed method exhibits a high degree of correlation with the designed waveform. The waveform matching error is evaluated quantitatively based on the switching loss difference between the simulated and designed waveforms. The error increases with the square of the maximum change in gate charge to achieve the designed waveform. Simulations indicate that the error can be suppressed by up to 5% when the maximum in gate charge change is within 15%.
As the simulation is performed with an ideal current source used to generate the gate current pulse for simplicity, it is worth demonstrating the method from an application perspective using a practical circuit. A switching waveform measurement system for a power MOSFET is constructed with a conventional voltage source gate driver and a simple gate current pulse generator consisting of operational amplifiers and transistors. The responses of drain-source voltage and drain current to small gate pulses are measured under the condition that the influence of the stray inductance of the power loop is negligible. The measured responses demonstrate that the prerequisites of the proposed method, which include the negligible delays in voltage and current response to the gate pulse and linear independence of all responses, are validated. The theoretical delay, given by the saturation electron drift velocity, is much smaller than the pulse widths generated in this study and those expected to be produced in actual applications involving high-speed switching power devices. The gate drive waveform is calculated using the obtained voltage and current responses to achieve the designed switching waveform. It is presented that the measured switching waveform of the MOSFET driven by the gate drive waveform is in good agreement with the designed. It is also demonstrated that the design of the drain-source voltage switching waveform can be conducted using the voltage responses to the gate charge pulses at any load current, as the load current values do not influence the responses.
In the aforementioned demonstration, the switching speed of the MOSFET is reduced to negate the influence of the stray inductance of the power loop. Experimental verification of the influence of stray inductance on the proposed method requires increasing the switching speed to induce a voltage across the inductance and narrowing the width of the small gate charge pulse to examine the voltage and current responses in the fast switching period. The linearity and bandwidth of the gate current pulse generator are improved, taking into account the requirements. Experimental outcomes show a good match between the obtained drain-source voltage waveform and the designed one. For the drain current responses to the small gate charge pulses, the voltage induced by the stray inductance results in interference with the adjacent pulses. This affects the accuracy in the design of the drain current waveform.
The proposed method creates a new strategy for open-loop control in active gate drives, as evidenced by its features compared with those of the method previously proposed. While the high performance of digital hardware is essential when the proposed method is implemented in practical applications, particularly in applications involving high-speed switching power devices, the usage of interleaving technology for digital-analog or analog-digital converters can mitigate the severe requirements in sampling rate and cost perspectives effectively. The performance requirements for digital hardware are estimated for a scenario using a power device in which switching is completed within 200nsec.
The novel method used to design switching waveforms presented in this thesis aids the identification of the optimal gate drive waveform in open-loop control schemes with digital active gate drives. Furthermore, it is conceivable that the concept of the switching waveform design achieved by the proposed method extends the useful application of active gate driving in power electronics.九州工業大学博士学位論文 学位記番号:生工博甲第493号 学位授与年月日:令和6年9月25日令和6年度doctoral thesi
Tetherbot: Experimental Demonstration and Path Planning of Cable-Driven Climbing in Microgravity
Get PDFIn this paper, we introduce Tetherbot, a cable-driven climbing robot designed for microgravity
environments with sparse holding points, such as space stations or asteroids. Tetherbot consists
of a platform with a robotic arm that is suspended via cables from multiple grippers. It achieves
climbing locomotion by alternately positioning the platform with the cables and relocating the grippers
with the robotic arm from one holding point to the next. The main contribution of this work is
the first experimental demonstration of autonomous cable-driven climbing in an environment with
sparse holding points. To this end, we outline the design, kinematics, and statics of the Tetherbot and
present a path planning algorithm to relocate the grippers. We demonstrate autonomous cable-driven
climbing through an experiment conducted in a simulated microgravity environment using the path
planning algorithm and a prototype of the robot. The results showcase Tetherbot’s ability to achieve
autonomous cable-driven climbing locomotion, thereby demonstrating that cable-driven climbing is
a viable concept and laying the foundation for future robots of this type.journal articl
Trimetal NiCoMn sulfides cooperated with two-dimensional Ti3C2 for high performance hybrid supercapacitor
Get PDFjournal articl
High-Density, Nonvolatile, Flexible Multilevel Organic Memristor Using Multilayered Polymer Semiconductors
Get PDFNonvolatile organic memristors have emerged as promising candidates for next-generation electronics, emphasizing the need for vertical device fabrication to attain a high density. Herein, we present a comprehensive investigation of high-performance organic memristors, fabricated in crossbar architecture with PTB7/Al-AlOx-nanocluster/PTB7 embedded between Al electrodes. PTB7 films were fabricated using the Unidirectional Floating Film Transfer Method, enabling independent uniform film fabrication in the Layer-by-Layer (LbL) configuration without disturbing underlying films. We examined the charge transport mechanism of our memristors using the Hubbard model highlighting the role of Al-AlOx-nanoclusters in switching-on the devices, due to the accumulation of bipolarons in the semiconducting layer. By varying the number of LbL films in the device architecture, the resistance of resistive states was systematically altered, enabling the fabrication of novel multilevel memristors. These multilevel devices exhibited excellent performance metrics, including enhanced memory density, high on–off ratio (>108), remarkable memory retention (>105 s), high endurance (87 on–off cycles), and rapid switching (∼100 ns). Furthermore, flexible memristors were fabricated, demonstrating consistent performance even under bending conditions, with a radius of 2.78 mm for >104 bending cycles. This study not only demonstrates the fundamental understanding of charge transport in organic memristors but also introduces novel device architectures with significant implications for high-density flexible applications.journal articl
Experimental Study on Erosion Process of Expressway Embankment Subjected to Tsunami After Earthquake
Get PDFOn March 11, 2011, the Great East Japan Earthquake triggered tsunamis that reached extensive areas along Japan’s Pacific coast. There have been instances where embankments built on plains for expressways mitigated the impact of tsunami damage. In the vicinity of the Sendai-tobu highway, the presence of an embankment approximately 10 m high altered the course of the advancing tsunami, thereby preventing flooding. Establishing a multiplied defense system using road embankments necessitates understanding the deformation and collapse mechanisms of road embankments impacted by tsunamis following seismic motion. In this study, overtopping experiments were conducted by first applying seismic motion to model embankments, followed by introducing the first wave of breaking bores, and then simulating prolonged overtopping by the tsunami. The experimental findings indicated that within the embankments impacted by the tsunami, there was an immediate increase in what is presumed to be pore air pressure following the arrival of the breaking bores, followed by a rise in pore water pressure during subsequent overtopping. Moreover, embankments subjected to seismic motion exhibited accelerated erosion following the overtopping. These results imply that when embankments settle due to an earthquake, leading to relatively higher anticipated inundation depths and the potential for overtopping, it is crucial to implement measures to prevent the settlement of the crest for embankments expected to serve as part of a multiplied defense system.journal articl
フードピックアンドプレースタスクにおけるロボティック認識向上のための合成データセットの生成
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483号 学位授与年月日:令和6年3月25日thesi
環境ビーコンを用いた屋内ローカライゼーションのためのリラベリングに基づくデータ補強
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488号 学位授与年月日:令和6年3月25日thesi
地理情報システムを利用した電力スポット市場価格予測システムの構築
Get PDF九州工業大学博士（工学）第1章 緒論| 第2章 電力市場とGISの概要| 第3章 GIS利用の説明変数の導出とスポット市場価格予測システムの構築| 第4章 最低価格の出現傾向の分析と価格予測への応用| 第5章 さらなる予測誤差低減に向けた検討| 第6章 予測の評価と既存手法との予測精度の比較| 第7章 結論近年、電力の自由化が進んでおり電力の取引の場として、電力市場の取引量が増大している。電力市場には先渡市場、スポット市場、時間前市場などの市場があり、それぞれの特徴に応じて取引が行われているが、取引量が最も多いのがスポット市場である。スポット市場は一日前市場とも呼ばれており、翌日の電気の取引が行われている市場である。スポット市場の価格は需要と供給のバランスで決定されるため、日時により価格が大きく異なるといった特徴がある。そのため、小売電気事業者においては入札前に価格の予測が可能であれば、安い時間帯に電力を購入し蓄電池等にため、高い時間帯に放電することで電力購入コストを低減することが可能である。
本研究において、スポット市場の価格予測を高精度に行うことを目的に研究をおこなった。スポット市場価格は需要と供給のバランスで決定されるため、価格予測システムの構築にあたり、需要と供給を詳細に説明できる説明変数の導出が必要である。筆者は需要と供給を説明する説明変数の一部が地理的な位置と関係があることに着目し、地理情報システム(GIS)を利用した価格予測システムの構築を行った。また、スポット市場価格予測システムの構築にあたり、ランニングコストを抑えることが必要となる。そのため、安価な気象予報とオープンデータの活用により、価格予測システムの構築を行った。
需要の要因としては、エアコンの消費電力量があげられる。エアコンの消費電力量は気温と関係を持つ一方、地域全体のエアコンの消費電力量を考慮する場合には人口も関係してくる。人口が多い地域ではエアコンの設置台数が多く、例えば夏場に気温が高いと消費電力量が大きくなり、地域全体の需要へ影響を及ぼす一方、人口が少ない地域ではこの影響が少ないためである。このエアコンの消費電力量を説明するために、GIS上で気温予報データと市町村ごとの人口データを重ね合わせることで「人口気温データ」の導出を行った。
供給の要因としては、近年導入量が増加している太陽光発電の発電量があげられる。特に電力需要の少ない春、秋においては太陽光発電の余剰によりスポット市場価格の下落が発生している。太陽光発電の発電量には、日射量が関係する一方、地域全体の太陽光発電の発電量を予測するためには、太陽光発電の導入量も関係してくる。そのため、GIS上で日射量予報と、市町村ごとの太陽光発電導入量を重ね合わせ、「予測太陽光発電量」の導出を行った。
太陽光発電の余剰と関連し、JEPXの最低取引価格である0.01円/kWhで電力が取引されることがある。これは、低需要期に太陽光発電の発電量が過剰な場合、供給過多に陥りこの最低価格で取引されるために発生する。筆者はこの最低価格の出現に着目し、最低価格の出現傾向を分析し、最低価格の出現傾向には、月の要素、曜日の要素、太陽光発電量の要素、出現傾向の要素があることを明らかにした。そのうえで、これ
らの要素を使用し「最低価格出現予測変数」の導出を行い、最低価格の出現を予測できる変数を導出した。
これらの三要素に加え、需要を説明する説明変数として日付と曜日データ、供給を説明する説明変数として、燃料価格と発電所の稼働状況を準備した。また、気象とスポット市場価格に関係があることに着目し雲量を、原子力発電所の稼働状況とスポット市場価格に関係があることに着目し、原子力発電所の稼働状況を加え価格予測システムの構築を行った。価格予測システムは、価格の予測を行う回帰モデルと、最低価格の出現を予測する分類モデルで構成されている。予測対象日前日までの説明変数とスポット市場価格でモデルの学習を行い、予測対象日の説明変数を利用することで予測を行うモデルとなっている。また、分類モデルが最低価格の出現を予測した場合、回帰モデルの出力を最低価格である0.01円/kWhとすることで最低価格の予測を可能としている。
本予測モデルで、九州エリアのスポット市場価格を対象に予測を行った。予測シミュレーションの結果、従来の手法よりも高精度に予測が可能であることを示すことができた。また、最低価格の出現を予測できるモデルであるため、小売電気事業者にとっては電気を購入するタイミングを検討しやすい予測モデルとなっている。また、高精度な気象予報であるMSMを使用した場合の価格予測シミュレーションも行い、誤差が低減されることも確認した。本予測モデルの利用により、小売電気事業者は蓄電池等との組み合わせで需要家に対し安価に電気を提供することができ、需要家の電力コストの抑制が期待される。また、発電事業者にとっても、売電する時間帯を蓄電池等との組み合わせで安い時間帯から価格の高い時間帯にずらすことで売電収益の向上が期待される。九州工業大学博士学位論文 学位記番号:工博甲第587号 学位授与年月日:令和6年3月25日令和5年度doctoral thesi
Temperature-Compensated Pure Silicon Cantilever Resonator with Coupled Torsional Structure at Anchor
Get PDFCombined with a coupled torsional structure at the anchor, a temperature compensated cantilever resonator was developed to reduce frequency dependence. The device shows the frequency dependence of 856 ppm with the temperature ranging from 25°C to 80°C. FEM simulation indicates that the coupled torsional structure at the anchor allows us to reduce temperature dependence of resonators to be 170 ppm with temperature ranging from −40°C to 80°C, optimizing wafer dopant, density, and mode.2023 22nd International Conference on Solid-State Sensors, Actuators and Microsystems (Transducers), 25-29 June 2023, Kyoto, Japanjournal articl