DSpace Collection:
http://hdl.handle.net/2115/20130
2024-03-28T09:49:04Zイジングマシンを用いた最適設計およびロバスト最適化に関する研究
http://hdl.handle.net/2115/91389
Title: イジングマシンを用いた最適設計およびロバスト最適化に関する研究
Authors: 丸尾, 昭人
Abstract: 社会のニーズの多様化・高度化に伴い、磁気デバイスの設計最適化問題はより高次元な問題として扱われるようになってきている。特に、設計変数が多くなり、膨大な組合せ問題が発生している。そのため、膨大な組合せ問題を効率的に解くことが求められている。そこで、組合せ最適化問題を解くことに特化した計算機であるイジングマシンに着目した。イジングマシンは二次制約なし二値最適化(Quadratic Unconstrained Binary Optimization, 以下 QUBO)に定式化された組合せ最適化問題を高速に解くことができる。したがって、デバイスの設計最適化問題を QUBO 形式に定式化をして、イジングマシンを用いて解くことが有効だと考えた。しかし、先行研究ではイジングマシンを用いて、磁気デバイスの設計最適化問題を解く手法は十分に議論されてきていなかった。また、実問題では、材料ばらつきや、組み立ての公差による寸法ばらつき、経時変化による物性変化などを起因とする不確定性により、システム・機器の特性が大きく損なわれることがある。そして、このような不確定性を考慮せずに最適化を行うと、変動に弱い解に収束することが考えられる。不確定性による変動に頑健な最適解を求めるためには、ロバスト最適化が必要となる。実問題では材料特性がばらつくことが多くあるため、材料ばらつきを考慮して形状を最適化することは重要なことである。先行研究では、モンテカルロ法や遺伝的アルゴリズムを用いて、ロバスト最適化を行う手法が提案されているが、前者は計算時間の問題、後者は高次元の場合にロバスト最適化の効果を十分に発揮できない問題が残されていた。上記の課題を踏まえ、本研究では、イジングマシンを用いた磁気デバイスの最適化法の開発および、材料ばらつき考慮のための材料特性の同定手法、材料ばらつきを考慮した設計最適化手法の開発を新たに検討した。まず、第 2 章では、イジングマシンを用いて磁気デバイスの設計最適化を扱うための基本となる磁石配列の最適化問題扱う。2 次元の磁石配列最適化問題を QUBO 形式に定式化し、イジングマシンを用いて磁石配列を最適化する。従来の磁石配列との比較や、最適化配列を振動発電デバイスに適用することで、提案手法の有効性を議論する。第 3 章では、第 2 章で提案したイジングマシンを用いた磁石配列最適化手法を、コイルと磁石配列の同時配置最適化手法に拡張する。提案手法を振動発電デバイスの最適化に適用することで、手法の有効性を議論する。第4 章では、第 2章で提案したイジングマシンを用いた磁石配列最適化手法を拡張し、磁石形状と磁性体形状のトポロジー最適化問題をイジングマシンを用いて最適化する手法を提案する。提案手法で得られた形状と参照形状との比較や、提案手法と従来の最適化手法の結果を比較することで、手法の有効性を議論する。第 5 章では、インダクタのインダクタンス特性からインダクタのコアの磁気特性を同定する新しい手法を提案する。提案手法を用いることで、特別な測定機器無しで、材料の BH 特性のための磁気ヒステリシス特性を得ることができる。提案手法では、測定された𝐿 − 𝐼特性から分布関数に含まれるパラメータを決定する。提案手法の同定結果と実測結果を比較することで手法の有効性を議論する。最後に、第 6 章では、共分散行列適応進化戦略を用いた磁気デバイスの新しいロバスト最適化手法を提案する。本手法では、計算量を増やすことなく、近傍個体の局所平均を用いて目的関数の期待値を評価する。第 5 章で求めた材料モデルを利用して、BH 特性のばらつきを考慮した磁気デバイスのトポロジー最適化に提案手法を適用し、従来手法と比較することで、手法の有効性を議論する。; With the diversification and sophistication of society's needs, the optimization of magnetic device design has become increasingly complex, often requiring the consideration of higher-dimensional problems. In particular, the number of design variables has increased, resulting in large-scale combinatorial optimization problems. Consequently, it is necessary to solve these optimization problems efficiently. Therefore, we focused on the Ising machine, a hardware system specialized for solving combinatorial optimization problems formulated as Quadratic Unconstrained Binary Optimization (QUBO) at high speed. However, previous studies have not sufficiently discussed how to solve the design optimization problem of magnetic devices using the Ising machine. Furthermore, in real-world problems, uncertainties caused by material variation, geometrical variation due to assembly tolerances, and changes in physical properties over time can significantly impair the characteristics of devices. Thus, robust optimization is required to obtain optimal solutions that are robust to fluctuations caused by these uncertainties. While prior research has proposed robust optimization methods utilizing Monte Carlo methods and robust genetic algorithms, the former is computationally time-consuming, and the latter is ineffective in high-dimensional cases. To overcome these problems, this thesis proposes a novel design optimization method for magnetic devices using the Ising machine and a robust design optimization method that accounts for material variation as follows:
・ Optimization of planar magnet array using Ising machines
・ Design optimization of coils and magnets using Ising machines
・ Topology optimization of magnetic devices using Ising machines
・ Identification method of material property
・ Robust design optimization method considering material variation2023-03-22T15:00:00Z丸尾, 昭人社会のニーズの多様化・高度化に伴い、磁気デバイスの設計最適化問題はより高次元な問題として扱われるようになってきている。特に、設計変数が多くなり、膨大な組合せ問題が発生している。そのため、膨大な組合せ問題を効率的に解くことが求められている。そこで、組合せ最適化問題を解くことに特化した計算機であるイジングマシンに着目した。イジングマシンは二次制約なし二値最適化(Quadratic Unconstrained Binary Optimization, 以下 QUBO)に定式化された組合せ最適化問題を高速に解くことができる。したがって、デバイスの設計最適化問題を QUBO 形式に定式化をして、イジングマシンを用いて解くことが有効だと考えた。しかし、先行研究ではイジングマシンを用いて、磁気デバイスの設計最適化問題を解く手法は十分に議論されてきていなかった。また、実問題では、材料ばらつきや、組み立ての公差による寸法ばらつき、経時変化による物性変化などを起因とする不確定性により、システム・機器の特性が大きく損なわれることがある。そして、このような不確定性を考慮せずに最適化を行うと、変動に弱い解に収束することが考えられる。不確定性による変動に頑健な最適解を求めるためには、ロバスト最適化が必要となる。実問題では材料特性がばらつくことが多くあるため、材料ばらつきを考慮して形状を最適化することは重要なことである。先行研究では、モンテカルロ法や遺伝的アルゴリズムを用いて、ロバスト最適化を行う手法が提案されているが、前者は計算時間の問題、後者は高次元の場合にロバスト最適化の効果を十分に発揮できない問題が残されていた。上記の課題を踏まえ、本研究では、イジングマシンを用いた磁気デバイスの最適化法の開発および、材料ばらつき考慮のための材料特性の同定手法、材料ばらつきを考慮した設計最適化手法の開発を新たに検討した。まず、第 2 章では、イジングマシンを用いて磁気デバイスの設計最適化を扱うための基本となる磁石配列の最適化問題扱う。2 次元の磁石配列最適化問題を QUBO 形式に定式化し、イジングマシンを用いて磁石配列を最適化する。従来の磁石配列との比較や、最適化配列を振動発電デバイスに適用することで、提案手法の有効性を議論する。第 3 章では、第 2 章で提案したイジングマシンを用いた磁石配列最適化手法を、コイルと磁石配列の同時配置最適化手法に拡張する。提案手法を振動発電デバイスの最適化に適用することで、手法の有効性を議論する。第4 章では、第 2章で提案したイジングマシンを用いた磁石配列最適化手法を拡張し、磁石形状と磁性体形状のトポロジー最適化問題をイジングマシンを用いて最適化する手法を提案する。提案手法で得られた形状と参照形状との比較や、提案手法と従来の最適化手法の結果を比較することで、手法の有効性を議論する。第 5 章では、インダクタのインダクタンス特性からインダクタのコアの磁気特性を同定する新しい手法を提案する。提案手法を用いることで、特別な測定機器無しで、材料の BH 特性のための磁気ヒステリシス特性を得ることができる。提案手法では、測定された𝐿 − 𝐼特性から分布関数に含まれるパラメータを決定する。提案手法の同定結果と実測結果を比較することで手法の有効性を議論する。最後に、第 6 章では、共分散行列適応進化戦略を用いた磁気デバイスの新しいロバスト最適化手法を提案する。本手法では、計算量を増やすことなく、近傍個体の局所平均を用いて目的関数の期待値を評価する。第 5 章で求めた材料モデルを利用して、BH 特性のばらつきを考慮した磁気デバイスのトポロジー最適化に提案手法を適用し、従来手法と比較することで、手法の有効性を議論する。
With the diversification and sophistication of society's needs, the optimization of magnetic device design has become increasingly complex, often requiring the consideration of higher-dimensional problems. In particular, the number of design variables has increased, resulting in large-scale combinatorial optimization problems. Consequently, it is necessary to solve these optimization problems efficiently. Therefore, we focused on the Ising machine, a hardware system specialized for solving combinatorial optimization problems formulated as Quadratic Unconstrained Binary Optimization (QUBO) at high speed. However, previous studies have not sufficiently discussed how to solve the design optimization problem of magnetic devices using the Ising machine. Furthermore, in real-world problems, uncertainties caused by material variation, geometrical variation due to assembly tolerances, and changes in physical properties over time can significantly impair the characteristics of devices. Thus, robust optimization is required to obtain optimal solutions that are robust to fluctuations caused by these uncertainties. While prior research has proposed robust optimization methods utilizing Monte Carlo methods and robust genetic algorithms, the former is computationally time-consuming, and the latter is ineffective in high-dimensional cases. To overcome these problems, this thesis proposes a novel design optimization method for magnetic devices using the Ising machine and a robust design optimization method that accounts for material variation as follows:
・ Optimization of planar magnet array using Ising machines
・ Design optimization of coils and magnets using Ising machines
・ Topology optimization of magnetic devices using Ising machines
・ Identification method of material property
・ Robust design optimization method considering material variationApplications of deep learning to accelerate label-free nerve imaging rate using coherent Raman rigid endoscopy : Construction of transfer learning method with fluorescence images and evaluation method for maximum imaging rate
http://hdl.handle.net/2115/91282
Title: Applications of deep learning to accelerate label-free nerve imaging rate using coherent Raman rigid endoscopy : Construction of transfer learning method with fluorescence images and evaluation method for maximum imaging rate
Authors: 大和, 尚記
Abstract: An imaging system for nerve visualization/identification during surgery is expected to improve the quality of life after surgery. This is because postsurgical pain and body dysfunction are induced by nerve damage during surgery. Although nerve-sparing approaches based on anatomical knowledge and surgeon’s experiments have been reported, it is unclear how well the peripheral nerves are preserved. The visibility of peripheral nerves is severely poor because peripheral nerves are transparent, colorless, and thin. Therefore, an intraoperative nerve dentification tool is required to improve the nerve-sparing rate and evaluate the efficiency of nerve-sparing approaches. In this thesis, an imaging acceleration system for a coherent anti-Stokes Raman scattering (CARS) rigid endoscope is developed for nerve-sparing surgery. The nerves can be visualized in a label-free manner using CARS, which is sensitive to species of molecular vibration in samples. The imaging rate acceleration of the CARS rigid endoscope developed in this laboratory is needed for clinical application. In this dissertation, the author shows the results of introducing image processing by deep learning to improve the nerve imaging rate of the CARS endoscope. The author developed the critical imaging rate (CIR) as a quantitative metric of imaging speed accelerated by deep learning. CIR is defined as the highest imaging rate to satisfy the image quality needed for medical images and is calculated from a harmonic mean of CIRPSNR and CIRSSIM which are the imaging rates satisfying criteria of PSNR=30 and SSIM=0.8, respectively. It is demonstrated that noise reduction of CARS endoscopic images made the imaging rate about five times faster using CIR (from 1.4 images/min to 7.0 images/min). Because noise reduction enhances the quality of the image, it is possible to improve the imaging rate of CARS rigid endoscopy by enhancing the quality of noisy nerve images captured at high rates. In general, a large dataset is needed for training deep-learning models. However, it is time-consuming and laborious to prepare a large dataset. The author proposed transfer learning using CARS microscopic images before training models on CARS endoscopic images. CARS microscopy made it easier to search peripheral nerves because the CARS images with a high signal-to-noise ratio can be obtained at a high imaging rate using an objective lens with a higher numerical aperture. The author compared three denoising models (WIN5R, DenoiseNet, and Noise2Noise) to determine the optimal model for denoising of CARS nerve images and to conduct pretraining before fine-tuning with CARS endoscopic images. Noise2Noise showed the highest denoising performance between the three models quantitatively and qualitatively. After training on CARS microscopic images, the model was fine-tuned using a few CARS endoscopic images. The author proposed a learning method to utilize images obtained with another modality as pre-training. Fluorescence microscopy was used as another modality to acquire a large dataset for nerve segmentation. Nerve segmentation is needed for displaying nerve areas to surgeons. The author prepared fluorescence images labeling lipids for pre-training nerve segmentation because CARS images have lipid information using CH2 vibration. U-Net, which is famous for semantic segmentation, was used. U-Net consists of an encoder part extracting feature maps from input images and a decoder part reconstructing output images from the feature maps. A VGG16 encoder, which learns to extract feature maps for image classification, was used in the encoder part of U-Net. It is found that both pre-training with fluorescence images and using the VGG16 encoder enhanced the segmentation performance significantly. In particular, pre-training with fluorescence images boosted the performance of the nerve segmentation from CARS images with a low signal-to-noise ratio. Therefore, the CIRsegmentation was improved about 47.5 times using semantic segmentation with deep learning than CIRdenoising. Semantic egmentation of nerves in CARS endoscopic images improved the imaging rate by about 47.5 times factor compared with denoising. These results open the opportunity for incorporating CARS rigid endoscopes into medical settings.2023-03-22T15:00:00Z大和, 尚記An imaging system for nerve visualization/identification during surgery is expected to improve the quality of life after surgery. This is because postsurgical pain and body dysfunction are induced by nerve damage during surgery. Although nerve-sparing approaches based on anatomical knowledge and surgeon’s experiments have been reported, it is unclear how well the peripheral nerves are preserved. The visibility of peripheral nerves is severely poor because peripheral nerves are transparent, colorless, and thin. Therefore, an intraoperative nerve dentification tool is required to improve the nerve-sparing rate and evaluate the efficiency of nerve-sparing approaches. In this thesis, an imaging acceleration system for a coherent anti-Stokes Raman scattering (CARS) rigid endoscope is developed for nerve-sparing surgery. The nerves can be visualized in a label-free manner using CARS, which is sensitive to species of molecular vibration in samples. The imaging rate acceleration of the CARS rigid endoscope developed in this laboratory is needed for clinical application. In this dissertation, the author shows the results of introducing image processing by deep learning to improve the nerve imaging rate of the CARS endoscope. The author developed the critical imaging rate (CIR) as a quantitative metric of imaging speed accelerated by deep learning. CIR is defined as the highest imaging rate to satisfy the image quality needed for medical images and is calculated from a harmonic mean of CIRPSNR and CIRSSIM which are the imaging rates satisfying criteria of PSNR=30 and SSIM=0.8, respectively. It is demonstrated that noise reduction of CARS endoscopic images made the imaging rate about five times faster using CIR (from 1.4 images/min to 7.0 images/min). Because noise reduction enhances the quality of the image, it is possible to improve the imaging rate of CARS rigid endoscopy by enhancing the quality of noisy nerve images captured at high rates. In general, a large dataset is needed for training deep-learning models. However, it is time-consuming and laborious to prepare a large dataset. The author proposed transfer learning using CARS microscopic images before training models on CARS endoscopic images. CARS microscopy made it easier to search peripheral nerves because the CARS images with a high signal-to-noise ratio can be obtained at a high imaging rate using an objective lens with a higher numerical aperture. The author compared three denoising models (WIN5R, DenoiseNet, and Noise2Noise) to determine the optimal model for denoising of CARS nerve images and to conduct pretraining before fine-tuning with CARS endoscopic images. Noise2Noise showed the highest denoising performance between the three models quantitatively and qualitatively. After training on CARS microscopic images, the model was fine-tuned using a few CARS endoscopic images. The author proposed a learning method to utilize images obtained with another modality as pre-training. Fluorescence microscopy was used as another modality to acquire a large dataset for nerve segmentation. Nerve segmentation is needed for displaying nerve areas to surgeons. The author prepared fluorescence images labeling lipids for pre-training nerve segmentation because CARS images have lipid information using CH2 vibration. U-Net, which is famous for semantic segmentation, was used. U-Net consists of an encoder part extracting feature maps from input images and a decoder part reconstructing output images from the feature maps. A VGG16 encoder, which learns to extract feature maps for image classification, was used in the encoder part of U-Net. It is found that both pre-training with fluorescence images and using the VGG16 encoder enhanced the segmentation performance significantly. In particular, pre-training with fluorescence images boosted the performance of the nerve segmentation from CARS images with a low signal-to-noise ratio. Therefore, the CIRsegmentation was improved about 47.5 times using semantic segmentation with deep learning than CIRdenoising. Semantic egmentation of nerves in CARS endoscopic images improved the imaging rate by about 47.5 times factor compared with denoising. These results open the opportunity for incorporating CARS rigid endoscopes into medical settings.Algal Blooms as Marine Ecosystem Risk: Forecasting Spread and Biogeochemical Stress
http://hdl.handle.net/2115/91245
Title: Algal Blooms as Marine Ecosystem Risk: Forecasting Spread and Biogeochemical Stress
Authors: 王, 浩炯
Abstract: Algae, accounting for less than one percent of Earth’s total photosynthetic biomass, are remarkable carbon drawdown contributors, fixing nearly half of the world’s organic carbon, especially during algal blooms. However, escalating concerns surround algal blooms, their persistence, and distribution, serving as indicators of both global climate shifts and local anthropogenic pressures. These phenomena intertwine with coastal marine ecosystems worldwide, where tide disruptions and human-induced disturbances increasingly degrade water quality, fostering frequent algal blooms. The growing duration of these blooms poses a significant threat, impacting vital ecological processes and services, such as carbon cycling and sequestration. Yet, unraveling the complex interplay between ecological factors and environmental stressors, along with deciphering algal bloom patterns, remains a formidable challenge due to limited data and a lack of universally applicable analytical approaches. An innovative predictive model merges transfer entropy network inference with a forecasting graph neural network to anticipate both blooming and non-blooming epidemic scenarios, along with their underlying environmental factors that elucidate bloom sources, causes and systemic risk. This model exhibits strong predictive capabilities, extracting crucial ecosystem features even in the absence of spatial dependencies. A novel 2D entropic ecosystem mandala is introduced, wherein the ecological impact, manifested through the distribution’s Cyanobacteria-driven chlorophylla (CHL-a) randomness, correlates proportionally with systemic environmental stress, governed by erratic oceanic, climatic, and coastal nutrient factors. Originally, a spatial risk was defined based on CHL-a magnitude, persistence and shifts. Through a case study in Florida Bay (FL Bay), we unveil how algal bloom shifts endure in shallow regions with elevated dinoflagellate-to-diatom ratios, underscoring Cyanobacteria’s pivotal role in phytoplankton dynamics and the influence of terrestrial discharge on marine microbiome equilibrium. This unfolding scenario presents formidable challenges, notably the heightened potential for green-blue algal blooms (associated with river dominance) to trigger harmful red tides, with cascading socio-ecological impacts spanning carbon cycling disruptions and entrenched eutrophication in coastal ecosystems. A universal threshold on the top 20% Pareto extremes of CHL-a, distinctly defines bloom and non-bloom phases, independent of endemic or epidemic categorization, driven by distinct eco-environmental interactions, where the paramount biogeochemical stress follows a scale-free structure with CHL-a acting as the central hub. Predicting algal blooms in the short and long term is crucial for assessing the well-being of ecosystems, encompassing coastal-marine environments, species, and human populations. Furthermore, it offers insights into the effects on environmental processes like carbon sequestration. However, the escalating disruption in biogeochemical balance compromises our capacity to forecast algal blooms, barring during outbreaks when intervention becomes belated. This deficiency hampers the investigation and management of the underlying eco-environmental factors triggering undesirable algal bloom occurrences and propagation. And our ideas improved this difficulty to some extent, both in terms of causal inference and model prediction.2023-12-24T15:00:00Z王, 浩炯Algae, accounting for less than one percent of Earth’s total photosynthetic biomass, are remarkable carbon drawdown contributors, fixing nearly half of the world’s organic carbon, especially during algal blooms. However, escalating concerns surround algal blooms, their persistence, and distribution, serving as indicators of both global climate shifts and local anthropogenic pressures. These phenomena intertwine with coastal marine ecosystems worldwide, where tide disruptions and human-induced disturbances increasingly degrade water quality, fostering frequent algal blooms. The growing duration of these blooms poses a significant threat, impacting vital ecological processes and services, such as carbon cycling and sequestration. Yet, unraveling the complex interplay between ecological factors and environmental stressors, along with deciphering algal bloom patterns, remains a formidable challenge due to limited data and a lack of universally applicable analytical approaches. An innovative predictive model merges transfer entropy network inference with a forecasting graph neural network to anticipate both blooming and non-blooming epidemic scenarios, along with their underlying environmental factors that elucidate bloom sources, causes and systemic risk. This model exhibits strong predictive capabilities, extracting crucial ecosystem features even in the absence of spatial dependencies. A novel 2D entropic ecosystem mandala is introduced, wherein the ecological impact, manifested through the distribution’s Cyanobacteria-driven chlorophylla (CHL-a) randomness, correlates proportionally with systemic environmental stress, governed by erratic oceanic, climatic, and coastal nutrient factors. Originally, a spatial risk was defined based on CHL-a magnitude, persistence and shifts. Through a case study in Florida Bay (FL Bay), we unveil how algal bloom shifts endure in shallow regions with elevated dinoflagellate-to-diatom ratios, underscoring Cyanobacteria’s pivotal role in phytoplankton dynamics and the influence of terrestrial discharge on marine microbiome equilibrium. This unfolding scenario presents formidable challenges, notably the heightened potential for green-blue algal blooms (associated with river dominance) to trigger harmful red tides, with cascading socio-ecological impacts spanning carbon cycling disruptions and entrenched eutrophication in coastal ecosystems. A universal threshold on the top 20% Pareto extremes of CHL-a, distinctly defines bloom and non-bloom phases, independent of endemic or epidemic categorization, driven by distinct eco-environmental interactions, where the paramount biogeochemical stress follows a scale-free structure with CHL-a acting as the central hub. Predicting algal blooms in the short and long term is crucial for assessing the well-being of ecosystems, encompassing coastal-marine environments, species, and human populations. Furthermore, it offers insights into the effects on environmental processes like carbon sequestration. However, the escalating disruption in biogeochemical balance compromises our capacity to forecast algal blooms, barring during outbreaks when intervention becomes belated. This deficiency hampers the investigation and management of the underlying eco-environmental factors triggering undesirable algal bloom occurrences and propagation. And our ideas improved this difficulty to some extent, both in terms of causal inference and model prediction.Algal Blooms as Marine Ecosystem Risk: Forecasting Spread and Biogeochemical Stress [an abstract of dissertation and a summary of dissertation review]
http://hdl.handle.net/2115/91238
Title: Algal Blooms as Marine Ecosystem Risk: Forecasting Spread and Biogeochemical Stress [an abstract of dissertation and a summary of dissertation review]
Authors: 王, 浩炯2023-12-24T15:00:00Z王, 浩炯A Study on the Design Optimization of the Bipolar Permanent Magnet Type Low-field MRI Device
http://hdl.handle.net/2115/91236
Title: A Study on the Design Optimization of the Bipolar Permanent Magnet Type Low-field MRI Device
Authors: 孔, 暁涵
Abstract: In recent years, portable low-field Magnetic resonance imaging (MRI) devices have been developed to complement high-field superconducting MRI. Portable low-field MRI devices offer advantages such as being lightweight, movable, and providing low-cost diagnostic services compared to the commonly used high-field MRI devices. However, there are still some challenges to be addressed, particularly concerning the electromagnetic (EM) structure, including gradient coil design and permanent magnets design. Based on how the main magnetic field is generated, there are different types of low-field MRI device. Among these, the bipolar permanent magnet type low-field MRI device is commonly used due to its advantages, such as good magnetic field homogeneity, structural compactness, and an open imaging area. However, some problems remain to be studied, especially about the EM structure including gradient coil design and permanent magnets design. In this paper, we focus on the design optimization of the bipolar permanent magnet type low-field MRI device, the main content of the thesis is as follows: In Chapter 1, the research background and motivations are introduced, and the contributions of this study are also summarized. In Chapter 2, a novel method for designing gradient coils for low-field MRI devices is proposed. The proposed method considers the effect of magnetic materials, particularly anti- eddy plates, by introducing image dipole currents. In the optimal design of gradient coils, the effect of ferromagnetic materials is minimized to obtain highly linear fields. The magnetic field measurement results and phantom images reveal the validity of the proposed method. In Chapter 3, a design method for Z-gradient coils in low-field MRI systems is proposed, focusing on enhancing anti-eddy performance. The newly introduced design procedure significantly improves the anti-eddy performance of the coils. Measurement and imaging results demonstrate that the optimal coil exhibited superior anti-eddy performance compared to conventional coils. In Chapter 4, a multi-fidelity topology optimization method has been proposed to alleviate the local optima problem. This method simplifies the design difficulty by dividing the optimization into sub-problems at the physical level. The proposed method shows a better performance than the conventional method in the design of low-field MRI devices. In Chapter 5, a passive shimming method is proposed for fine-tuning the static magnetic field in a low-field MRI device. A test case validated the effectiveness of this approach, reducing non-uniformity from 10,000 ppm to 125 ppm after three iterations. In Chapter 6, conclusions and future works are discussed.2023-12-24T15:00:00Z孔, 暁涵In recent years, portable low-field Magnetic resonance imaging (MRI) devices have been developed to complement high-field superconducting MRI. Portable low-field MRI devices offer advantages such as being lightweight, movable, and providing low-cost diagnostic services compared to the commonly used high-field MRI devices. However, there are still some challenges to be addressed, particularly concerning the electromagnetic (EM) structure, including gradient coil design and permanent magnets design. Based on how the main magnetic field is generated, there are different types of low-field MRI device. Among these, the bipolar permanent magnet type low-field MRI device is commonly used due to its advantages, such as good magnetic field homogeneity, structural compactness, and an open imaging area. However, some problems remain to be studied, especially about the EM structure including gradient coil design and permanent magnets design. In this paper, we focus on the design optimization of the bipolar permanent magnet type low-field MRI device, the main content of the thesis is as follows: In Chapter 1, the research background and motivations are introduced, and the contributions of this study are also summarized. In Chapter 2, a novel method for designing gradient coils for low-field MRI devices is proposed. The proposed method considers the effect of magnetic materials, particularly anti- eddy plates, by introducing image dipole currents. In the optimal design of gradient coils, the effect of ferromagnetic materials is minimized to obtain highly linear fields. The magnetic field measurement results and phantom images reveal the validity of the proposed method. In Chapter 3, a design method for Z-gradient coils in low-field MRI systems is proposed, focusing on enhancing anti-eddy performance. The newly introduced design procedure significantly improves the anti-eddy performance of the coils. Measurement and imaging results demonstrate that the optimal coil exhibited superior anti-eddy performance compared to conventional coils. In Chapter 4, a multi-fidelity topology optimization method has been proposed to alleviate the local optima problem. This method simplifies the design difficulty by dividing the optimization into sub-problems at the physical level. The proposed method shows a better performance than the conventional method in the design of low-field MRI devices. In Chapter 5, a passive shimming method is proposed for fine-tuning the static magnetic field in a low-field MRI device. A test case validated the effectiveness of this approach, reducing non-uniformity from 10,000 ppm to 125 ppm after three iterations. In Chapter 6, conclusions and future works are discussed.A Study on the Design Optimization of the Bipolar Permanent Magnet Type Low-field MRI Device [an abstract of dissertation and a summary of dissertation review]
http://hdl.handle.net/2115/91230
Title: A Study on the Design Optimization of the Bipolar Permanent Magnet Type Low-field MRI Device [an abstract of dissertation and a summary of dissertation review]
Authors: 孔, 暁涵2023-12-24T15:00:00Z孔, 暁涵Estimation of Hosting Capacity of Photovoltaic Generations in Distribution Networks using Hybrid Particle Swarm and Gradient Descent Optimization
http://hdl.handle.net/2115/91229
Title: Estimation of Hosting Capacity of Photovoltaic Generations in Distribution Networks using Hybrid Particle Swarm and Gradient Descent Optimization
Authors: Zulu, Esau
Abstract: The excessive dependence on fossil fuels such as coal, oil and gas for energy production has led to massive emission of CO 2 . This huge emission of CO 2 in the atmosphere has led to deterioration of the ozone layer. The subsequent impact of this has been rapid global temperature rise and, ultimately, climate change. To avoid further deterioration of the ozone layer and avoid deepening the climate change crisis, the world has, over the last few decades, resorted to the use of clean green-energy resources such as wind, photovoltaic (PV) etc., for the world’s energy needs. In the same vein, electrical vehicles (EV) with battery energy storage systems (BESS) have increased in the share of the transportation industry to replace fossil fuel dependent transportation. PV power sources have been increasingly adopted in large quantities and accounts for nearly ninety percent of green-energy power sources in the electrical power distribution networks (DN). This is because PV is relatively easy to install, has higher scalability and is cheaper than other renewable energy options. However, the adoption of PV in huge quantities can lead to various challenges in the operation of the distribution networks. The greatest challenge posed by PV is the risk of over-voltages during times of high solar irradiation (with subsequent high- power output) at times of low power demand. Other risks include, thermal capitulation of network lines and cables, reverse power flows, and high harmonics. Therefore, there is a need to determine the amount of PV power which a particular DN can accommodate without abrogating the network’s operational limits. This amount is referred to as the PV hosting capacity (PVHC). This study proposes an efficient method for estimating the PVHC of a DN. This method uses swarm intelligence in combination with gradient descent. The method harnesses the excellent exploration capabilities of particle swarm optimization (PSO) and the powerful exploitation of the optimum solution espoused by the gradient descent algorithm. In hybridizing the PSO and the GD algorithms, the proposed method also gets rid of the ills of each method. The proposed method’s efficacy in depth and speed of calculation was tested on several DN test systems including the IEEE 33 bus test DN, the IEEE 69 test DN and the existing 136 bus in Sao Paulo, Brazil to estimate the PVHC of these networks. The proposed method was also used in the study of the effects of BESS and EV on the PVHC of a DN. The results of the calculations were compared with several other methods. The numerical results of the simulations proved that the proposed method was more efficient compared with other methods found in literature. The study also proposes the use of the deterministic approach in combination with the stochastic methods to produce a fast optimization algorithm for estimating the PV hosting capacity distribution networks operating under the uncertainties which inherent in the network variables. In this part of the research, the PSO-GD was combined with the PEM-based probabilistic load flow analysis to synthesize a powerful tool for estimating the acceptable limit of PV which can be safely installed into the distribution network without violating the network performance limits. This tool can be used for network planning purposes at the conception stage of the DN or for system expansion planning purposes.2023-12-24T15:00:00ZZulu, EsauThe excessive dependence on fossil fuels such as coal, oil and gas for energy production has led to massive emission of CO 2 . This huge emission of CO 2 in the atmosphere has led to deterioration of the ozone layer. The subsequent impact of this has been rapid global temperature rise and, ultimately, climate change. To avoid further deterioration of the ozone layer and avoid deepening the climate change crisis, the world has, over the last few decades, resorted to the use of clean green-energy resources such as wind, photovoltaic (PV) etc., for the world’s energy needs. In the same vein, electrical vehicles (EV) with battery energy storage systems (BESS) have increased in the share of the transportation industry to replace fossil fuel dependent transportation. PV power sources have been increasingly adopted in large quantities and accounts for nearly ninety percent of green-energy power sources in the electrical power distribution networks (DN). This is because PV is relatively easy to install, has higher scalability and is cheaper than other renewable energy options. However, the adoption of PV in huge quantities can lead to various challenges in the operation of the distribution networks. The greatest challenge posed by PV is the risk of over-voltages during times of high solar irradiation (with subsequent high- power output) at times of low power demand. Other risks include, thermal capitulation of network lines and cables, reverse power flows, and high harmonics. Therefore, there is a need to determine the amount of PV power which a particular DN can accommodate without abrogating the network’s operational limits. This amount is referred to as the PV hosting capacity (PVHC). This study proposes an efficient method for estimating the PVHC of a DN. This method uses swarm intelligence in combination with gradient descent. The method harnesses the excellent exploration capabilities of particle swarm optimization (PSO) and the powerful exploitation of the optimum solution espoused by the gradient descent algorithm. In hybridizing the PSO and the GD algorithms, the proposed method also gets rid of the ills of each method. The proposed method’s efficacy in depth and speed of calculation was tested on several DN test systems including the IEEE 33 bus test DN, the IEEE 69 test DN and the existing 136 bus in Sao Paulo, Brazil to estimate the PVHC of these networks. The proposed method was also used in the study of the effects of BESS and EV on the PVHC of a DN. The results of the calculations were compared with several other methods. The numerical results of the simulations proved that the proposed method was more efficient compared with other methods found in literature. The study also proposes the use of the deterministic approach in combination with the stochastic methods to produce a fast optimization algorithm for estimating the PV hosting capacity distribution networks operating under the uncertainties which inherent in the network variables. In this part of the research, the PSO-GD was combined with the PEM-based probabilistic load flow analysis to synthesize a powerful tool for estimating the acceptable limit of PV which can be safely installed into the distribution network without violating the network performance limits. This tool can be used for network planning purposes at the conception stage of the DN or for system expansion planning purposes.Estimation of Hosting Capacity of Photovoltaic Generations in Distribution Networks using Hybrid Particle Swarm and Gradient Descent Optimization [an abstract of dissertation and a summary of dissertation review]
http://hdl.handle.net/2115/91225
Title: Estimation of Hosting Capacity of Photovoltaic Generations in Distribution Networks using Hybrid Particle Swarm and Gradient Descent Optimization [an abstract of dissertation and a summary of dissertation review]
Authors: Zulu, Esau2023-12-24T15:00:00ZZulu, EsauUnique Surface Enhanced Raman Scattering Induced by Plasmon-Nanocavity Coupling and its Application to Elucidating the Mechanism of Enhanced Water Oxidation Under the Strong Coupling Conditions
http://hdl.handle.net/2115/91216
Title: Unique Surface Enhanced Raman Scattering Induced by Plasmon-Nanocavity Coupling and its Application to Elucidating the Mechanism of Enhanced Water Oxidation Under the Strong Coupling Conditions
Authors: Zang, Xiaoqian
Abstract: Noble metal nanoparticles (NPs) such as gold and silver, are of great interest due to their unique optical, magnetic, and electronic properties based on the plasmon resonance. The localized surface plasmon resonance (LSPR) is the collective oscillation of the conduction band electrons at the surface of metal NPs which induces significant electromagnetic field enhancement at the metal NPs surface. Recently, it has been reported that the modal coupling between an LSPR and a Fabry-Pérot (FP) nanocavity mode can enhance the photochemical reactions taking place near the metal NPs. It is great interesting and critical to investigate the enhancement mechanism of plasmon-induced photoelectrochemical (PEC) reaction under the plasmon-nanocavity coupling. In this thesis, the effect of plasmon-nanocavity coupling on the near-field distribution was investigated by a measure of surface-enhanced Raman scattering (SERS). Besides, the plasmon-nanocavity coupling structure was applied to investigate the plasmon-induced water oxidation reaction, which affected by the plasmon-nanocavity coupling, using in situ electrochemical surface-enhanced Raman scattering (EC-SERS) measurements. A plasmon-nanocavity coupling structure consist of Au NPs/TiO2/Au-film (ATA) was fabricated to investigate the spatial coherence effect by SERS measurements. Compared to the Au NPs/TiO2 (AT) structures without FP nanocavity, the SERS signal collected on ATA was enhanced by 11 times because of the dramatic near-field enhancement causing by the coupling between LSPR of Au NPs and FP nanocavity resonance. Besides the large near-field enhancement, a spatially homogeneous near-field intensity was observed on ATA, which can be attributed to the coherent coupling between the LSPR of each Au NP and the FP nanocavity. Simulations also showed the homogeneous near-field distribution under the plasmon-nanocavity coherent coupling, which supports our experiment observations (Chapter 2). To investigate the effect of plasmon-nanocavity coupling on the plasmon-induced water oxidation reaction, an Au-Ag alloy NPs/TiO2/Au-film (AATA) structure was employed, and in situ EC-SERS measurements were performed to detect the intermediate species of plasmon-induced water oxidation. The Au-Ag alloy NPs were deposited on TiO2/Au-film to create a modal strong coupling between the LSPR of Au-Ag alloy NPs and the FP nanocavity resonance. A large splitting energy was observed on the AATA structure which was derived from the large oscillator strength of the LSPR of Au-Ag alloy NPs. The Raman intensity of the Au-O and Au-OH stretching vibrations, which are characterized intermediate species of the plasmon-induced water oxidation on Au-based NPs, were systematically studied at a wide range of electrochemical potentials. Compared with Au-Ag alloy NPs/TiO2 (AAT) structure without FP nanocavity, the in situ EC-SERS measurement of the intermediate species on AATA electrode showed higher sensitivity. More interestingly, the Raman signals on AATA showed a more negative onset potential than the AAT structures, indicating a much more efficient charge separation on AATA structures that facilitates water oxidation reaction. This enhanced water oxidation efficiency on AATA is likely attributed to the quantum coherence between the Au-Ag alloy NPs through the nanocavity, leading to the accumulation of a large number of holes (Chapter 3). In summary, the near-field intensity distribution, and the water oxidation reaction intermediate on the plasmon-nanocavity coherent coupling structure were investigated by means of SERS measurements. The SERS measurements revealed a large near-field enhancement and spatially homogeneous near- field distribution under the plasmon-nanocavity coherent coupling. Furthermore, the intermediates of plasmon-induced water oxidation were investigated using the plasmon-nanocavity coherent coupling structure by in situ EC-SERS measurement. From the EC-SERS measurements, a more negative onset potential of the water oxidation intermediates was observed on AATA, which was attributed the higher near-field enhancement and the efficient plasmon-induced charge separation in the coherent aera under the plasmon-nanocavity coherent coupling.2023-12-24T15:00:00ZZang, XiaoqianNoble metal nanoparticles (NPs) such as gold and silver, are of great interest due to their unique optical, magnetic, and electronic properties based on the plasmon resonance. The localized surface plasmon resonance (LSPR) is the collective oscillation of the conduction band electrons at the surface of metal NPs which induces significant electromagnetic field enhancement at the metal NPs surface. Recently, it has been reported that the modal coupling between an LSPR and a Fabry-Pérot (FP) nanocavity mode can enhance the photochemical reactions taking place near the metal NPs. It is great interesting and critical to investigate the enhancement mechanism of plasmon-induced photoelectrochemical (PEC) reaction under the plasmon-nanocavity coupling. In this thesis, the effect of plasmon-nanocavity coupling on the near-field distribution was investigated by a measure of surface-enhanced Raman scattering (SERS). Besides, the plasmon-nanocavity coupling structure was applied to investigate the plasmon-induced water oxidation reaction, which affected by the plasmon-nanocavity coupling, using in situ electrochemical surface-enhanced Raman scattering (EC-SERS) measurements. A plasmon-nanocavity coupling structure consist of Au NPs/TiO2/Au-film (ATA) was fabricated to investigate the spatial coherence effect by SERS measurements. Compared to the Au NPs/TiO2 (AT) structures without FP nanocavity, the SERS signal collected on ATA was enhanced by 11 times because of the dramatic near-field enhancement causing by the coupling between LSPR of Au NPs and FP nanocavity resonance. Besides the large near-field enhancement, a spatially homogeneous near-field intensity was observed on ATA, which can be attributed to the coherent coupling between the LSPR of each Au NP and the FP nanocavity. Simulations also showed the homogeneous near-field distribution under the plasmon-nanocavity coherent coupling, which supports our experiment observations (Chapter 2). To investigate the effect of plasmon-nanocavity coupling on the plasmon-induced water oxidation reaction, an Au-Ag alloy NPs/TiO2/Au-film (AATA) structure was employed, and in situ EC-SERS measurements were performed to detect the intermediate species of plasmon-induced water oxidation. The Au-Ag alloy NPs were deposited on TiO2/Au-film to create a modal strong coupling between the LSPR of Au-Ag alloy NPs and the FP nanocavity resonance. A large splitting energy was observed on the AATA structure which was derived from the large oscillator strength of the LSPR of Au-Ag alloy NPs. The Raman intensity of the Au-O and Au-OH stretching vibrations, which are characterized intermediate species of the plasmon-induced water oxidation on Au-based NPs, were systematically studied at a wide range of electrochemical potentials. Compared with Au-Ag alloy NPs/TiO2 (AAT) structure without FP nanocavity, the in situ EC-SERS measurement of the intermediate species on AATA electrode showed higher sensitivity. More interestingly, the Raman signals on AATA showed a more negative onset potential than the AAT structures, indicating a much more efficient charge separation on AATA structures that facilitates water oxidation reaction. This enhanced water oxidation efficiency on AATA is likely attributed to the quantum coherence between the Au-Ag alloy NPs through the nanocavity, leading to the accumulation of a large number of holes (Chapter 3). In summary, the near-field intensity distribution, and the water oxidation reaction intermediate on the plasmon-nanocavity coherent coupling structure were investigated by means of SERS measurements. The SERS measurements revealed a large near-field enhancement and spatially homogeneous near- field distribution under the plasmon-nanocavity coherent coupling. Furthermore, the intermediates of plasmon-induced water oxidation were investigated using the plasmon-nanocavity coherent coupling structure by in situ EC-SERS measurement. From the EC-SERS measurements, a more negative onset potential of the water oxidation intermediates was observed on AATA, which was attributed the higher near-field enhancement and the efficient plasmon-induced charge separation in the coherent aera under the plasmon-nanocavity coherent coupling.Unique Surface Enhanced Raman Scattering Induced by Plasmon-Nanocavity Coupling and its Application to Elucidating the Mechanism of Enhanced Water Oxidation Under the Strong Coupling Conditions [an abstract of dissertation and a summary of dissertation review]
http://hdl.handle.net/2115/91213
Title: Unique Surface Enhanced Raman Scattering Induced by Plasmon-Nanocavity Coupling and its Application to Elucidating the Mechanism of Enhanced Water Oxidation Under the Strong Coupling Conditions [an abstract of dissertation and a summary of dissertation review]
Authors: Zang, Xiaoqian2023-12-24T15:00:00ZZang, Xiaoqian