2004 Fiscal Year Final Research Report Summary
Development of unified schemes that consist of modeling through controller design using a prior information model
| Project/Area Number |
14550451
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Control engineering
|
|---|
| Research Institution | OKAYAMA UNIVERSITY (2003-2004)
Osaka Prefecture University (2002) |
|---|
Principal Investigator |
IMAI Jun Okayama Univ., Electrical & Electronic Eng.Dept., Lecturer, 工学部, 講師 (50243986)
|
|---|
| Project Period (FY) |
2002 – 2004
|
| Keywords | A priori information model / modeling / controller design / distributed parameter systems / spectral analysis |
|---|
| Research Abstract |
Concerning the development of unified schemes that consists of modeling through controller design using a priori information model, we have obtained the following results :
1.A novel Auto Regressive(AR) model parameter estimation method has been proposed, which can utilize a prior information as well as time series data, by extending the Burg method on the basis of the Minimum Cross Entropy(MCE) principle. The proposed method has been applied to a spectral estimation of real speech signal and the estimated results have illustrated improved performances in comparison to those obtained by the Burg method.
2.For flexible structures represented by spatially vibrating distributed parameter systems, unstructured uncertainty caused by perturbation of physical parameters has been characterized as a certain shape in complex plane. Laboratory experiments using flexible beam illustrate feasibility of the controller design based on the characterization. These results demonstrate possibility of further development of gray-box modeling scheme that shrinks feasible set of systems based on frequency response data.
3.A regulator problem for a heat conduction system, of which the index is made of the spatio integral of the squared deviation from reference temperature distribution. It is shown that through characterizing frequency response from input to temperature at each spatial point, a distributed parameter system with nominal model and additive uncertainty weight, both of which are real rational, is reconstructed using knowledge of the eigenstructure. The formulated problem is proved to be reduced to a standard mixed H 2/H_infinity one for a lincar finite dimensional time-invariant system. Numerical study demonstrates feasibility of the proposed design scheme.
|
