| Project/Area Number |
08650517
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計測・制御工学
|
|---|
| Research Institution | Waseda University |
|---|
Principal Investigator |
AKIZUKI Kageo Waseda Univ., Professor School of Sci.& Eng.,, 理工学部, 教授 (10063603)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
OURA Kunihiko Waseda Univ., Faculty Member Adv.Res.Inst.Sci.& Eng.,, 理工学総合研究センター, 講師 (40277819)
HANAZAKI Izumi Tokyo Denki Univ., Associate Professor, 理工学部, 助教授 (50180914)
|
|---|
| Project Period (FY) |
1996 – 1997
|
| Project Status |
Completed (Fiscal Year 1997)
|
| Budget Amount *help |
¥1,100,000 (Direct Cost: ¥1,100,000)
Fiscal Year 1997: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 1996: ¥600,000 (Direct Cost: ¥600,000)
|
|---|
| Keywords | Practical Plant / Closed-Loop Identification / Order and Delay-Time Estimation / 4SID method / Recurrent Neural Network / Fuzzy Model / ファジィ制御系 / システム同定 / 閉ループ系 / プラント次数 / むだ時間 / ニューロ / ファジィ / 異常診断 |
|---|
| Research Abstract |
This project treated several types of system identification from the point of their practical applications. To estimate available model to its purpose, which may be control, prediction or diagnosis, one should select its model structure appropriately. We mainly considered system identification in the case when the plant was controlled by feedback loop, or the plant was nonlinear system. We used linear time series model, neuro model and fuzzy model as identification models, and investigated the method to make appropriate models from the data. Some procedures for estimating orders and delay-time were proposed by using input and output data. Research results are summarized as follows :
1.We proposed a system identification procedure useful for practical data observed by closed-loop experiments. It includes several steps, aiming for model selection and order estimation. Simulation experiment showed usefulness of the procedure.
2.We applied 4SID (Subspace based State-Space System IDentification) method to order estimation, and showed usefulness of the method even if the plant is noisy, by using comparatively long data.
3.We used recurrent neural network for system identification and proposed a line to construct an appropriate neuro model for the data.
4.We proposed a procedure to construct a fuzzy control system, typically on the selection of model structures by considering the influence of noisy data.
|
