Development and Organical Coordination of Practical Identification Algorithms for Large Scale Systems
Grant-in-Aid for General Scientific Research (C)
|Allocation Type||Single-year Grants |
|Research Institution||Kyushu University |
SAGARA Setsuo Kyushu Univ., Faculty of Engg., Professor, 工学部, 教授 (60037679)
OBAYASHI Masanao Kyushu Univ., Faculty of Engg, Research Associate, 工学部, 助手 (60213849)
DANJYO Mitsuaki Niihama National College of Tech., Dept. of Electrical Engg., Assistant Professo, 講師 (70032026)
MURATA Junichi Kyushu Univ., Faculty of Engg, Associate Professor, 工学部, 助教授 (60190914)
WADA Kiyoshi Kyushu Univ., Faculty of Engg, Associate Professor, 工学部, 助教授 (60125127)
KUMAMARU Kousuke Kyushu Inst. of Tech., Faculty of Computer Science and Systems Engg., Professor, 情報工学部, 教授 (30037949)
三木 康臣 九州大学, 工学部, 助手 (90209734)
|Project Period (FY)
1987 – 1989
Completed (Fiscal Year 1989)
|Budget Amount *help
¥1,900,000 (Direct Cost: ¥1,900,000)
Fiscal Year 1989: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 1988: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 1987: ¥900,000 (Direct Cost: ¥900,000)
|Keywords||System Identification / Parameter Estimation / State Estimation / Multivariable System / Fault Diagnosis / Distributed Parameter System / Power System Control / Load Forecasting / ファジィ推論 / 有限次元近似 / 多変数線形システム / 未知パタン変動システム / 分布定数系 / 電力系統 / 同定 / 未知境界条件の推定|
The following systems have been studied and the results are summarized below: (1) multivariable systems, (2) systems with unknown pattern changes, (3) distributed parameter systems and (4) power systems.
(1) A computationally efficient algorithm was proposed for obtaining a generalized least squares estimate and a bias compensated least squares method (LSM) was extended to the noisy input-output measurement case. The numerically unstable phenomena in LSM were studied in relation to system and noise characteristics. Overlapping form was proposed to be an effective system description among vector difference equations for multivariable system identification.
(2) Faults in dynamical systems are represented by unexpected pattern changes. To detect them a two-level hierarchical diagnosing system was proposed and an advance assessing method was devised for fault detectabilities. Moreover two specialized diagnosing systems were developed: one uses fuzzy inference to cope with difficulties in ada
ptive control systems; the other is subsystem-coordinator system for decentralized systems.
(3) Approximation of a distributed parameter system to a lumped parameter system was studied with a focus on the coordinate functions in the Galerkin method. As to parameter estimation in partial differential equations, practical methods were developed using numerical integrating filter and optimal digital filter. An industrially practical problem of indirect measurement of thermal characteristics of boiling water at hot metal surface was treated as a boundary condition estimation problem and an efficient method was proposed.
(4) Load forecasting problems have been studied as modeling problems of complex systems. Two types of modeling methods have been developed: automatic model selection for one-day-ahead load forecasting and time series modeling for weekly forecasting. A new algorithm was proposed to obtain a output feedback controller for power system generators. Though the controller has a simple structure, it has considerably better outputs responses and robustness property compared with traditional one. Less
Report (4 results)
Research Products (35 results)