• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Development of neural networks based on physical models and exploration of meteorological physics

Research Project

Project/Area Number 19K22876
Research Category

Grant-in-Aid for Challenging Research (Exploratory)

Allocation TypeMulti-year Fund
Review Section Medium-sized Section 61:Human informatics and related fields
Research InstitutionOsaka University

Principal Investigator

Fukui Ken-ichi  大阪大学, 産業科学研究所, 准教授 (80418772)

Co-Investigator(Kenkyū-buntansha) 冨田 智彦  熊本大学, 大学院先端科学研究部(理), 准教授 (20344301)
Project Period (FY) 2019-06-28 – 2021-03-31
Project Status Completed (Fiscal Year 2020)
Budget Amount *help
¥6,370,000 (Direct Cost: ¥4,900,000、Indirect Cost: ¥1,470,000)
Fiscal Year 2020: ¥2,730,000 (Direct Cost: ¥2,100,000、Indirect Cost: ¥630,000)
Fiscal Year 2019: ¥3,640,000 (Direct Cost: ¥2,800,000、Indirect Cost: ¥840,000)
Keywordsディープラーニング / 偏微分方程式 / 温度風 / マルチタスク学習 / 敵対的サンプル生成 / 物理モデル / PINN / 気象予測 / 残差項 / 温度風方程式 / 地衡風 / データ駆動 / モデル駆動 / モデル残差
Outline of Research at the Start

本研究は,人工ニューラルネットワークにおいて運動方程式などの既知物理モデルによる制約を導入することで,物理モデルに従う成分とそのモデル残差項を分解して学習可能な新規フレームワークの確立を目的とする.機械学習によるデータ駆動型予測と物理モデルによるモデル駆動型予測の融合を図ることで,両者の欠点を補い合い予測精度の向上が期待できる.予測精度のみでなく,残差成分を分解して予測可能になることで学理探究への糸口となる.本研究では,気象における対流圏上層の風予測を題材に,予測精度の向上とデータ駆動型科学として気象学の進展を目指す.

Outline of Final Research Achievements

In this work, we tackled the following two researches. Frist, we proposed a deep learning architecture that can decompose and output known physical model component and model residual component. Then, we verified the accuracy of the proposed method with the task of estimating the wind velocity in the upper troposphere from the atmospheric conditions in the lower layer. The validity of the wind vector distribution of the residual component by the proposed method was justified from the knowledge of meteorology. Second, we proposed an improvement method by multi-task learning and adversarial exsample generation for the method of obtaining the value of the solution at an arbitrary position of the partial differential equation, by automatic differentiation and deep learning. We confirmed the improvement of estimation accuracy for some basic PDEs.

Academic Significance and Societal Importance of the Research Achievements

本研究成果の1. 物理モデル成分とモデル残差成分を分解して出力可能なディープラーニングアーキテクチャと,2. ディープラーニングと自動微分による偏微分方程式の求解の高精度化,の両者を将来的に統合することで,End-to-Endで任意の観測量から偏微分方程式を満足するように任意の物理量を推定することが可能になる.本統合方式により,観測データと既知の方程式を有する自然科学において,両者を活用して予測精度の向上や最適化など広く応用が期待できる.

Report

(3 results)
  • 2020 Annual Research Report   Final Research Report ( PDF )
  • 2019 Research-status Report
  • Research Products

    (10 results)

All 2021 2020 2019

All Journal Article (1 results) (of which Peer Reviewed: 1 results) Presentation (9 results) (of which Int'l Joint Research: 3 results,  Invited: 1 results)

  • [Journal Article] Physics-Guided Neural Network with Model Discrepancy Based on Upper Troposphere Wind Prediction2019

    • Author(s)
      Ken-ichi Fukui, Junya Tanaka, Tomohiko Tomita, and Masayuki Numao
    • Journal Title

      Proc. 2019 18th IEEE International Conference on Machine Learning and Applications (ICMLA)

      Volume: n/a Pages: 414-419

    • DOI

      10.1109/icmla.2019.00078

    • Related Report
      2019 Research-status Report
    • Peer Reviewed
  • [Presentation] Adversarial Multi-task Learning Algorithm for Solving Partial Differential Equations2021

    • Author(s)
      Pongpisit Thanasutives, Masayuki Numao, Ken-ichi Fukui
    • Organizer
      Japan Geoscience Union Meeting 2021
    • Related Report
      2020 Annual Research Report
  • [Presentation] Learning to Solve Multiple Partial Differential Equations Using Physics-informed Neural Networks2021

    • Author(s)
      Pongpisit Thanasutives, Masayuki Numao, Ken-ichi Fukui
    • Organizer
      2021年度人工知能学会全国大会(第35回)
    • Related Report
      2020 Annual Research Report
  • [Presentation] Adversarial Multi-task Learning Enhanced Physics-informed Neural Networks for Solving Partial Differential Equations2021

    • Author(s)
      Pongpisit Thanasutives, Masayuki Numao and Ken-ichi Fukui
    • Organizer
      2021 The International Joint Conference on Neural Networks (IJCNN2021)
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research
  • [Presentation] A Periodic Convolutional Recurrent Network Model for Climate Prediction2020

    • Author(s)
      Ekasit Phermphoonphiphat, Tomohiko Tomita, Masayuki Numao, Ken-ichi Fukui
    • Organizer
      2020年度人工知能学会全国大会(第34回)
    • Related Report
      2020 Annual Research Report
  • [Presentation] Physics-Guided Neural Network with Model Discrepancy Based on Upper Troposphere Wind Prediction2019

    • Author(s)
      Ken-ichi Fukui, Junya Tanaka, Tomohiko Tomita, and Masayuki Numao
    • Organizer
      2019 18th IEEE International Conference on Machine Learning and Applications (ICMLA)
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] データ駆動とモデル駆動の融合によるディープラーニングと気象予測2019

    • Author(s)
      福井健一
    • Organizer
      IT連携フォーラムOACIS第36回シンポジウム
    • Related Report
      2019 Research-status Report
    • Invited
  • [Presentation] Study of Upper Tropospheric Circulations over the Northern Hemisphere Prediction Using Multivariate Features by ConvLSTM2019

    • Author(s)
      Ekasit Phermphoonphiphat, Tomohiko Tomita, Masayuki Numao, and Ken-ichi Fukui
    • Organizer
      The 23nd Asia Pacific Symposium on Intelligent and Evolutionary Systems (IES2019)
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] A Prediction of Upper Tropospheric Circulations over the Northern Hemisphere Using ConvLSTM2019

    • Author(s)
      Ekasit Phermphoonphiphat, Tomohiko Tomita, Masayuki Numao, and Ken-ichi Fukui
    • Organizer
      電子情報通信学会人工知能と知識処理研究会
    • Related Report
      2019 Research-status Report
  • [Presentation] Spatiotemporal Climate Forecasting with ConvLSTM2019

    • Author(s)
      Ekasit Phermphoonphiphat, Tomohiko Tomita, Masayuki Numao, and Ken-ichi Fukui
    • Organizer
      日本地球惑星科学連合2019年大会
    • Related Report
      2019 Research-status Report

URL: 

Published: 2019-07-04   Modified: 2022-01-27  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi