| 研究課題/領域番号 |
22K14590
|
|---|
| 研究種目 |
若手研究
|
| 配分区分 | 基金 |
|---|
| 審査区分 |
小区分29010:応用物性関連
|
|---|
| 研究機関 | 東京理科大学 |
|---|
研究代表者 |
|
|---|
| 研究期間 (年度) |
2022-04-01 – 2025-03-31
|
| 研究課題ステータス |
交付 (2023年度)
|
| 配分額 *注記 |
4,680千円 (直接経費: 3,600千円、間接経費: 1,080千円)
2024年度: 780千円 (直接経費: 600千円、間接経費: 180千円)
2023年度: 780千円 (直接経費: 600千円、間接経費: 180千円)
2022年度: 3,120千円 (直接経費: 2,400千円、間接経費: 720千円)
|
|---|
| キーワード | Magnetostriction / TDA / Machine learning / Magnetization / Magnetic domains / Persistent homology / Machine Learning |
|---|
| 研究開始時の研究の概要 |
Sensors are important devices in DX because they connect the real to the virtual world. Ferromagnetic shape memory alloy (FSMA) has been an intense research subject as actuators of sensors due to its high magnetostriction coefficient, low cost, and high-speed drive capability. The main topic of this research is to write a machine-learning outputted energy formula that describes the magneto-mechanical properties in real material based on structural and morphological properties to understand and improve FSMAs.
|
| 研究実績の概要 |
The research aims to explore the magnetostriction mechanism using topological data analysis and machine learning. By incorporating the material's morphology, we seek to develop a new, comprehensive theory.
Last year, we published a paper titled "Visualization of the Magnetostriction Mechanism in Fe-Ga Alloy Single Crystal Using Dimensionality Reduction Techniques" (doi: 10.1109/TMAG.2023.3312372) and presented our findings at the IEEE International Magnetics Conference.
This year, our focus has been on simulating ferromagnetic shape memory alloys (FSMA) by integrating metallography and magnetic properties. We developed custom software that combines phase-field and micromagnetics simulations. Using a mix of topological data analysis (persistent homology) and Fast Fourier Transformation (FFT), we described the the effect of stress through energy landscapes. I conducted simulations that integrate phase-field methods with micromagnetics, employing topological data analysis and unsupervised learning. This helped me identify key latent features in magnetization images and link them to their physical meanings. I observed energy barriers corresponding to changes in the magnetic domain structure of FSMA and visualized the energy exchange among different energy terms. This approach has allowed me to correlate microstructures with macro properties through free energy analysis.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
3: やや遅れている
理由
The primary goal for this year was to fully develop a comprehensive simulation that integrates metallography and mechanical properties. This simulation uses phase-field methods to simulate the martensite transformation and micromagnetics simulations based on the Landau-Lifshitz-Gilbert (LLG) equation.
To date, we have made significant progress in developing this simulation and conducting preliminary analyses.
The initial analysis, which combines persistent homology and Fast Fourier Transformation (FFT), has shown promising results. It indicates that the formation of energy barriers can be explained by varying the stress within the material. However, further in-depth analysis is required to fully elucidate the underlying mechanisms. Our next steps involve refining the simulation and conducting more detailed analyses to achieve a comprehensive understanding of how stress influences the formation of energy barriers and, ultimately, the magnetostriction mechanism.
|
| 今後の研究の推進方策 |
Last year, the objective was to enhance the simulation to seamlessly integrate strain stress with metallography and magnetic properties. In the first year, I analyzed Magneto-Optical Kerr Effect (MOKE) microscope images and used feature extraction techniques and machine learning to determine their contribution to magnetostriction. The second year was dedicated to understanding the fundamental physical principles involved.In the previous year, we focused on the simulation to gain a comprehensive understanding of the system.
This year, the goal is to study how latent features influence local energy and how to control these interactions. Our observations indicated that analyzing energy contributions at individual regions, rather than the total average, provided interesting insights into the mechanism. We plan to use persistent homology and inverse analysis to link important regions contributing to magnetostriction with specific energy contributions. Achieving this connection will clarify the mechanism and allow us to incorporate metallographic information into the theoretical framework. This approach will enable a deeper understanding of how individual regions contribute to the overall energy landscape, thus providing a clearer picture of the magnetostriction mechanism.
|
