Adaptive Hierarchical Network Model Reduction of Large-Scale Systems Strictly Preserving Finite Frequency Properties

2015 Fiscal Year Final Research Report

• Project/Area Number: 25820177
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Control engineering/System engineering
• Research Institution: The University of Tokyo
• Principal Investigator: KOJIMA CHIAKI 東京大学, 情報理工学(系)研究科, 助教 (00456162)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 制御理論 / 階層化ネットワークシステム / モデル低次元化 / 固有直交分解 / 有限周波数特性 / 安定性診断 / 分散カルマンフィルタ / 電力ネットワーク

Outline of Final Research Achievements

In this research project, we first derived a model reduction method preserving dissipativity which is stronger property than finite frequency properties. Next, we proposed a distributed approach to the proper orthogonal decomposition which is one of efficient model reduction method of large-scale systems, and theoretically guaranteed an upper bound of an approximation error. Based on this approach, we also derived a method for transient stability diagnosis of power networks. In addition to these results, we characterized finite frequency properties of multi-dimensional systems in terms of dissipation inequality, which can be regarded as an original result. Moreover, we clarified an estimation performance of a distributed Kalman filter with diffusion strategies. This can be an efficient result for a design of the filter.

Free Research Field

システム制御工学