Loop decomposition and dynamic interaction analysis of decentralized control systems

https://doi.org/10.1016/0009-2509(95)00386-XGet rights and content

Abstract

Based on a rigorous formulation and analysis, a decentralized control system is structurally decomposed into separate SISO control loops with interactions unmasked. In particular, the decomposition provides a completely equivalent representation of the original multivariable system, and consequently recovers some important dynamic information that is often ignored in most of the existing dynamic analysis methods and interaction measures. Absolute and relative interactions in individual loops are defined and measured and their structural significance is clearly shown. This allows for the effects of interaction on a closed-loop system to be evaluated in a transparent way. This article provides an overview of the usefulness of the proposed decomposition approach and interaction measurement. The evaluation of the severity of loop coupling and performance deviation from independent design are discussed. Possible difficulties imposed by interactions, effects of controller tuning and variable pairing on closed-loop performance are also dealt with.

References (27)

  • J.P. Gagnepain et al.

    Analysis of process interactions with applications to multiloop control systems design

    Ind. Engng Chem. Process Des. Dev.

    (1982)
  • P. Grosdidier et al.

    Closed-loop properties from steady-state gain information

    Ind. Engng Chem. Fundam.

    (1985)
  • N. Jensen et al.

    Interaction analysis in multivariable control system

    A.I.Ch.E. J.

    (1986)
  • Cited by (19)

    • Quantification of interaction in multiloop control systems using directed spectral decomposition

      2013, Automatica
      Citation Excerpt :

      Consequently, several methods have been developed to incorporate the dynamic nature of the control system into the analysis of interaction (Grosdidier & Morari, 1987; He et al., 2009; Hovd & Skogestad, 1992; Jensen, Fisher, & Shah, 1986). The existing methods can broadly be classified into two categories, (i) those which assume a perfect controller (Bristol, 1978; Gagnepain & Seborg, 1982; Witcher & McAvoy, 1977) and (ii) those which take into account the actual controller (Grosdidier & Morari, 1987; Huang, Oshima, & Hashimoto, 1994; Meeuse & Huesman, 2002; Zhu & Jutan, 1996). An overview of several categories of interaction analysis methods from an input–output pairing perspective is provided in van de Wal and de Jager (2001).

    • New methods for interaction analysis of complex processes using weighted graphs

      2012, Journal of Process Control
      Citation Excerpt :

      Nevertheless, this impact is very low (0.076) compared with the effect on y2 (0.924), and it is also existing only at high frequencies as it can be appreciated in the second column of FDPTc(Ey · ΩCL) in Fig. 11 (right). The same conclusions were obtained in [30] using loop decomposition, and they were validated by examining the responses to independent setpoint changes in both control loops. These step responses are depicted in Fig. 12.

    • Design of multiple FACTS controllers for damping inter-area oscillations: A decentralised control approach

      2004, International Journal of Electrical Power and Energy System
      Citation Excerpt :

      It follows that control action in the ui−yi loop in response to changes and/or disturbances in this control loop initiates control action in one or more control loops to keep the variables, yl, l≠i at their set points. This problem is known as interaction and may result in a degraded overall performance for the MIMO system [14,15]. In the equations above, gii(s) represents the interaction-free transfer function of the system, and δi(s) represents the additional dynamics in the ith loop resulting from other control loops.

    View all citing articles on Scopus
    View full text