2023 Fiscal Year Final Research Report
A systematic framework from modeling to control for systems including stochastic uncertainty based on statistical models and stochastic control systems theory
Project/Area Number |
21K04106
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Review Section |
Basic Section 21040:Control and system engineering-related
|
Research Institution | Osaka University |
Principal Investigator |
Satoh Satoshi 大阪大学, 大学院工学研究科, 教授 (60533643)
|
Project Period (FY) |
2021-04-01 – 2024-03-31
|
Keywords | ガウス過程回帰システム / 離散時間確率システム / 確率サンプル値制御 |
Outline of Final Research Achievements |
This study established a method that can handle the entire process from modeling to control for systems including stochastic uncertainty, based on stochastic systems control theory and the Gaussian process regression model. There have been no conventional control methods that stochastically consider the variance of Gaussian process regression models. Thus, we formulated discrete-time stochastic systems that correspond to the statistical features of Gaussian process regression models. Using this correspondence, we developed a systematic framework for modeling, analyzing, and controlling the Gaussian process regression models based on stochastic systems control theory. Furthermore, we proposed a control methodology guaranteeing the stability of the behavior between sample points of the training data based on stochastic sampled-data control theory and performed experimental verification.
|
Free Research Field |
制御工学
|
Academic Significance and Societal Importance of the Research Achievements |
①ガウス過程回帰により得られるシステムと統計的対応を有する確率システムを定式化したことで,これまでに培われた確率システム制御理論に基づく系統的な解析・制御が可能となった. ②実験データからのモデル表現において利点をもつ一方で制御系設計を困難にしていたガウス過程回帰システムに対して,モデル化から制御までの一貫した解析・制御器設計を可能とする枠組みを構築し,実機検証による有効性の確認も行った. ③実システムは連続時間システムであっても,実験データの取得は離散的となるため,ガウス過程回帰により得られるモデルは離散時間システムとなるが,本研究によりサンプル点間の挙動の安定性も保証した制御法を与えた.
|