1998 Fiscal Year Final Research Report Summary
Intelligent control of nonlinear systems based upon case-based reasoning
| Project/Area Number |
08455197
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計測・制御工学
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
MAEDA Hajime Graduate School of Engineering, Professor, 大学院・工学研究科, 教授 (60029535)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
ITO Yoshimichi Graduate School of Engineering, As-sistant Professor, 大学院・工学研究科, 助手 (10263203)
IIGUNI Youji Graduate School of Engineering, As-sociate Professor, 大学院・工学研究科, 助教授 (80168054)
|
|---|
| Project Period (FY) |
1996 – 1998
|
| Keywords | nonlinear system / wavelet transform / database / feature extraction / nonlinear signal processing / control law |
|---|
| Research Abstract |
An approach to intelligent control and signal processing based upon case-based reasoning is presented. The wavelet transform is used to extract I/O characteristics of nonlinear plant, and the multiresolution pattern analysis method is applied to distinguish global and local features of the I/O signals. Using a hierarchical image coding technique, the extracted feasters are compressed without any degradation and their efficient storage procedure is designed. A nonlinear adaptive estimation method based on the n-nearest neighbor approach is derived to estimate the output for a query from the I/O data seen so far. Observed I/O data are stored in a database in the form of a k-d trie and a nonlinear local model to answer each query is derived based upon regularization theory. An efficient time-updating scheme of the database contents to follow nonstationary data is derived. A storage procedure allowing a simple and efficient update is developed for reduction in processing time and storage requirement. The effectiveness of the proposed method is demonstrated with both simulation data and real speech signals. Furthermore, an automatic landing system is developed on the basis of a human skill model. The model is expressed as a nonlinear I/O mapping from the aircraft state to the control command provided by a human expert. All the I/O data provided by the human expert are stored in a database a priori, and a local model to answer the current aircraft state is built to generate a control input close to the human operation. A gain adaptation technique is introduced for improving the robustness.
|
