2022 Fiscal Year Final Research Report
Improvement of the method to estimate physical property distribution by fusion of physical model and machine learning
Project/Area Number |
20K15219
|
Research Category |
Grant-in-Aid for Early-Career Scientists
|
Allocation Type | Multi-year Fund |
Review Section |
Basic Section 31020:Earth resource engineering, Energy sciences-related
|
Research Institution | Kyoto University |
Principal Investigator |
|
Project Period (FY) |
2020-04-01 – 2023-03-31
|
Keywords | ベイズ統計学 / 岩石物性 / 熱水システム / 空隙率 / 機械学習 |
Outline of Final Research Achievements |
In the Earth resource development, the estimations of subsurface physical properties and their distribution are important on the evaluation of the potentials of the resource and the selection of drilling locations. This study developed machine learning-based methods to estimate physical properties of the target area, such as porosities. The developed methods were examined by applying it to a geothermal field. The study further developed methods for measuring rocks to improve the quality and quantity of data, allowing the developed machine learning-based methods to be applied to promising areas of earth resources.
|
Free Research Field |
地球資源工学
|
Academic Significance and Societal Importance of the Research Achievements |
地球資源開発において、空隙率等の地下の物性値の推定は不可欠である。本研究で開発した手法は、機械学習と物理モデルを組み合わせることによって、厳密な物理モデルが不明瞭な条件下でも一定程度の物理的な妥当性を考慮して、物性値を推定できるフレームワークを構築した。さらに、推定値のばらつきの評価を行い、どの範囲の推定値まで信頼がおけるか評価を可能とした。また、応力や弾性波速度、熱伝導率といった観測値をより効率的に取得する手法を開発し、従来よりも多くの地域や条件でデータ取得が可能となることが期待される。
|