| Project/Area Number |
16560385
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Control engineering
|
|---|
| Research Institution | The University of Tokushima |
|---|
Principal Investigator |
IKEDA Kenji The University of Tokushima, Institute of Technology and Science, Associate Professor, ソシオテクノサイエンス研究部, 助教授 (80232180)
|
|---|
| Project Period (FY) |
2004 – 2006
|
| Project Status |
Completed (Fiscal Year 2006)
|
| Budget Amount *help |
¥3,300,000 (Direct Cost: ¥3,300,000)
Fiscal Year 2006: ¥1,000,000 (Direct Cost: ¥1,000,000)
Fiscal Year 2005: ¥1,100,000 (Direct Cost: ¥1,100,000)
Fiscal Year 2004: ¥1,200,000 (Direct Cost: ¥1,200,000)
|
|---|
| Keywords | Adaptive observer / Continuous-time model / Recursive Least Squares Method / Sampled I / O data / State variable filters / Asymptotic bias / 連続時間モデル / 逐次最小2乗法 / サンプル値データシステム / 初期状態推定 / 過渡応答 |
|---|
| Research Abstract |
First, a basic structure of an identification method by using adaptive observer is proposed, in which a continuous-time model of the plant is estimated from a finite length of sampled I/O data. There is an information loss incurred by the sampling, intersample output of the plant is estimated as well as the plant parameters. The plant parameters is used for the estimation of the intersample output. Therefore, the proposed method becomes an iterative algorithm.
Next, discrete time realization of the state variable filters is investigated, while the state variable filters in the previous studies are realized by continuous time filters. The discretization of the state variable filters gives us a good insight into the followings :
(1) Output of the state variable filters becomes a pseudo-linear regression vector instead of a linear regression vector, because there is a mechanism to estimate the intersample output.
(2) The proposed method to iterate the recursive least squares method corresponds to a bootstrap method.
(3) Under the assumptions on the persistent excitation of the pseudo-linear regression vector and the stability of the state variable filters, the proposed method becomes a contraction map and the existence of the limit of the estimated parameters and the exponential convergence of the parameter is proved under a certain conditions.
(4) An upper bound of the parameter estimation error is proportional to the standard deviation of the noise.
Finally, the problem is reformulated in the stochastic framework instead of the deterministic framework and the stochastic property of the estimated parameter error is analyzed. Based on this analysis, a more efficient and consistent method of parameter estimation is proposed.
|
