| Project/Area Number |
18K13966
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 25030:Disaster prevention engineering-related
|
|---|
| Research Institution | University of Yamanashi |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2018-04-01 – 2023-03-31
|
| Project Status |
Completed (Fiscal Year 2022)
|
| Budget Amount *help |
¥4,030,000 (Direct Cost: ¥3,100,000、Indirect Cost: ¥930,000)
Fiscal Year 2021: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2020: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2019: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2018: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
|
|---|
| Keywords | リアルタイム地震被害推定 / 深層学習 / マルチモーダル学習 / 機械学習と物理の統合 / リモートセンシング / 災害被害検知 / データサイエンス / 物理とデータ駆動のハイブリッド / 地震被害判別 / 都市ビッグデータ / 地震被害検知 / AI / 衛星画像 / マルチモーダル機械学習 / 衛星リモートセンシング / 機械学習 / 異常検知 / 不均衡データ / 地震被害推定 |
|---|
| Outline of Final Research Achievements |
In order to estimate damage information on residential structures in real-time, which is important for appropriate disaster response in the event of a major earthquake, machine learning methods have been developed to analyze data contributing to the estimation of earthquake damage distribution from various information sources in an integrated manner.
A multimodal deep learning model was developed to analyze the prior information from the structure type and age and the observation information from satellite images, and a Bayesian framework was developed to integrate the damage estimation results of the deep learning model with those from physical simulations, and its effectiveness was verified. Simultaneously, through these studies, new academic trends in the hybridization and integration of machine learning methods and physical analysis methods are discussed.
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究により，90%超という従来の技術水準を超える高い精度での住宅の倒壊有無の検知に成功する手法が開発され，発災直後の初動期の対応の変革に繋がりうる技術の確立に寄与する成果が得られた．また，本研究を通じて得られた，機械学習技術と物理的な解析技術の統合手法に関する知見は，地震被害推定に留まらない重要な学術的動向を指摘するものである．
|
