A graph-theoretic method to find decentralized fixed modes of LTI systems

Department of Control and Dynamical Systems, California Institute of Technology, Pasadena, CA 91125, USA; Department of Electrical and Computer Engineering, Concordia University, Montréal, Que., Canada H3G 1M8
Automatica 01/2007; DOI: 10.1016/j.automatica.2007.04.019
Source: DBLP

ABSTRACT This paper deals with the decentralized pole assignability of interconnected systems by means of linear time-invariant (LTI) controllers. A simple graph-theoretic approach is proposed to identify the distinct decentralized fixed modes (DFMs) of the system, i.e., the unrepeated modes which cannot be moved by means of a LTI decentralized controller. The state-space representation of the system is transformed to the decoupled form using a proper change of coordinates. For any unrepeated mode, a matrix is then computed which resembles the transfer function matrix of the system at some point in the complex plane. A bipartite graph is constructed accordingly in terms of the computed matrix. Now, the problem of verifying if this mode is a DFM of the system reduces to checking if the constructed graph has a complete bipartite subgraph with a certain property. The sole restriction of this work is that it is only capable of identifying the distinct DFMs of a system. However, it is axiomatic that most of the modes of the real-world systems are normally distinct. The primary advantage of the present paper is its simplicity, compared to the existing ones which often require evaluating the rank of several matrices.

  • [Show abstract] [Hide abstract]
    ABSTRACT: Cited By (since 1996): 1, Export Date: 15 November 2012, Source: Scopus, doi: 10.1108/17506221111120901, Language of Original Document: English, Correspondence Address: Doukas, H.; Management and Decision Support Systems Lab (EPU-NTUA), School of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece; email:, References: Awerbuch, S., (2000) Getting it right: The real cost impacts of a renewables portfolio standard, ,, available at: (accessed January 2009);
    International Journal of Energy Sector Management 01/2011; 5(1):60-80.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Given a centralized controller associated with a linear time-invariant interconnected system, this paper is concerned with designing a parameterized decentralized controller such that the state and input of the system under the obtained decentralized controller can become arbitrarily close to those of the system under the given centralized controller, by tuning the controller's parameters. To this end, a two-level decentralized controller is designed, where the upper level captures the dynamics of the centralized closed-loop system, and the lower level is an observed-based sub-controller designed based on the new notion of structural initial value observability. The proposed method can decentralize every generic centralized controller, provided the interconnected system satisfies very mild conditions. The efficacy of this work is elucidated by some numerical examples.
    IEEE Transactions on Automatic Control 01/2012; · 2.72 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: A relative degree matrix-based design method is proposed in this article by using graph theoretical methods for optimal control structure design consisting of single input–single output control loops. This approach enables us to use an efficient algorithm for finding a maximum weighted matching to solve the controller structure selection problem. The resulting optimal structures have been refined by analysing the zero dynamics of the input–output pairs. A novel method for decentralised controller structure retrofit is also proposed in this article. A graph theoretic algorithm is developed, which is based on finding the closest weighted maximum matching. Possible changes in efficiency in the controller structure, which may arise after retrofit, are illustrated using a heat exchanger network retrofit case study.
    International Journal of Systems Science 01/2011; 42:1023-1033. · 1.31 Impact Factor