| Project/Area Number |
14550447
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Control engineering
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
UOSAKI Katsuji Osaka University, Graduate School of Information Science and Technology, Professor, 大学院・情報科学研究科, 教授 (20029151)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HATANAKA Toshiharu Osaka University, Graduate School of Information Science and technology, Research Associate, 大学院・情報科学研究科, 助手 (10252884)
|
|---|
| Project Period (FY) |
2002 – 2004
|
| Project Status |
Completed (Fiscal Year 2004)
|
| Budget Amount *help |
¥3,400,000 (Direct Cost: ¥3,400,000)
Fiscal Year 2004: ¥900,000 (Direct Cost: ¥900,000)
Fiscal Year 2003: ¥1,000,000 (Direct Cost: ¥1,000,000)
Fiscal Year 2002: ¥1,500,000 (Direct Cost: ¥1,500,000)
|
|---|
| Keywords | nonlinear systems / system identification / evolutionary computation / local modeling / regime selection / block oriented models / RBF network / particle filter / 非線形モデル / ブロック指向モデル / 進化計算 / パラメータ推定 / 進化的計算 / 多目的最適化 / 非線形時系列 / 進化型計算法 / ウィナーモデル / ハンマーシュタインモデル |
|---|
| Research Abstract |
Since there exist nonlinearities in natal, social and industrial systems, nonlinear system modeling is ubiquitous for analysis, design, control, and optimization of such systems. In this we have developed new methodologies for nonlinear system modeling based on the idea of local modeling and evolutionary computation. The topics investigated are :
1. Automatic operating regime selection in local modeling
Since the performance of the global model is highly affected by the choice of the local operating regimes in which the local models are identified, automatic selection algorithms of suitable local regimes have been developed.
2. Identification of block oriented nonlinear models
Using evolutionary computation approaches, identification method for block oriented nonlinear models, in which nonlinear static part and linear dynamic part are in cascade connection.
3. Nonlinear modeling by Radial basis function network
Identification of radial basis function network for nonlinear modeling has been discussed from the multi-objective optimization problem.
4. Evolution strategies based particle filter for nonlinear filtering
A novel nonlinear filter called evolution strategies based nonlinear filter has been developed combining particle filter based on the importance filtering in Monte Carlo method and evolution strategies in evolutionary computation.
|
