Generalization Capability of Radial Basis Function Controller Using Random Search Method with Variable Search Length in Universal Learning Network
-
- Shao Ning
- Department of Electrical and Electronic Systems Engineering, Faculty of Information Science and Electrical Engineering, Kyushu University
-
- Hirasawa Kotaro
- Department of Electrical and Electronic Systems Engineering, Faculty of Information Science and Electrical Engineering, Kyushu University
-
- Ohbayashi Masanao
- Department of Electrical and Electronic Systems Engineering, Faculty of Information Science and Electrical Engineering, Kyushu University
-
- Togo Kazuyuki
- Department of Energy Conversion Engineering, Interdisciplinary Graduate School of Engineering Sciences, Kyushu University
この論文をさがす
抄録
In this paper, generalization capability of a Radial Basis Function controller using RasVal in Universal Learning Network was studied. RasVal is an abbreviation of Random Search with Variable Search Length and it can search for a global minimum systematically and effectively in a single framework which is not a combination of different methods. In this paper, a new method to overcome the over-fitting problem in nonlinear control systems is proposed, where the weighting coefficients of control variables in the criterion function are increased in order to obtain the generalization capability of RasVal. From simulation results of a nonlinear crane system, it has been shown that the smaller the scale of the R.B.F. controller is, the smaller the weighting coefficients of the control variables could be.
収録刊行物
-
- 九州大学大学院システム情報科学紀要
-
九州大学大学院システム情報科学紀要 2 (1), 31-37, 1997-03-26
九州大学大学院システム情報科学研究院
- Tweet
キーワード
詳細情報 詳細情報について
-
- CRID
- 1390009224843418752
-
- NII論文ID
- 110000579823
-
- NII書誌ID
- AN10569524
-
- DOI
- 10.15017/1475361
-
- ISSN
- 21880891
- 13423819
-
- HANDLE
- 2324/1475361
-
- 本文言語コード
- en
-
- データソース種別
-
- JaLC
- IRDB
- CiNii Articles
-
- 抄録ライセンスフラグ
- 使用可