| Project/Area Number |
17K06950
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Aerospace engineering
|
|---|
| Research Institution | Nihon University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
齊藤 允教 日本大学, 理工学部, 准教授 (20801020)
|
|---|
| Project Period (FY) |
2017-04-01 – 2022-03-31
|
| Project Status |
Completed (Fiscal Year 2021)
|
| Budget Amount *help |
¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2019: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2018: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2017: ¥2,860,000 (Direct Cost: ¥2,200,000、Indirect Cost: ¥660,000)
|
|---|
| Keywords | 振動燃焼 / ロケットエンジン / ガソリンエンジン / 冷炎 / 深層学習 / 自己符号化器 / 非線形力学系 / モード解析 / 燃焼振動 / 液滴燃焼 / モード分解 / 深層自己符号化器 / ロケット / トリプルフレーム / Deep Auto-encoder / 燃焼 / 振動解析 / ニューラルネットワーク |
|---|
| Outline of Final Research Achievements |
Deep learning is applied to the analysis of combustion dynamics which is the key for the safety and efficiency of combustors. This is to develop and validate the method for state estimation and for the determination of controlling phenomena in complex systems. The analyzed phenomena are the intrinsic combustion oscillation by the coupling between pressure and heat release fluctuations, and the intrinsic cool-flame oscillation where temperature and species concentration are coupled. The big data of physical quantities by numerical simulation of transient combustion field are input to the analysis. The phase and the energy fraction of oscillation modes are derived, as well as validating the ability of correlation analysis and near-term prediction. In addition, the controlling phenomena and location and phase at which they arise are clarified through the determination of the distinctive feature of combustion field.
|
| Academic Significance and Societal Importance of the Research Achievements |
燃焼ダイナミクスは非線形力学系で表される複雑現象で，従来の解析手法である直交モード分解では多数の独立モードが発現し，現象分析に高次元の位相空間が必要なことから理解困難な現象であった．これに対し深層学習で位相空間を張る変数を導くと，複雑な燃焼振動場の状態を２次元という低次元位相空間に表せ，さらにモードとの関係も求められるので，どのような現象がどこでどの順番で生じるかを容易に判別可能となった．冷炎ダイナミクス解析にも適用できたことから，例えば同じ連鎖分岐爆発問題であるパンデミックの感染者数の時間的空間的振動なども解析できると考えられ，一般の非線形力学系問題の解析法として応用が期待できる成果を得た．
|
