• DocumentCode
    2765689
  • Title

    Combined fault monitoring detection and control

  • Author

    Grimble, M.J.

  • Author_Institution
    Ind. Control Unit, Strathclyde Univ., Glasgow, UK
  • Volume
    4
  • fYear
    1998
  • fDate
    16-18 Dec 1998
  • Firstpage
    3675
  • Abstract
    One of the advantages of polynomial methods is the insights the frequency domain approach provides. The relatively new subject of combined fault monitoring and control has mostly been considered in a state-equation form but is it more difficult to determine the frequency domain characteristics. A new solution of the combined fault estimation and control problem is presented in the z-domain and the resulting robustness and reliability benefits are identified. An H2 cost-function is introduced that involves traditionally LQG cost terms and in addition fault estimation error components. The combined controller and fault estimator is economical to implement. The application of this methodology in ship roll stabilization problems is considered and the benefits of the approach are identified. The presentation is illustrated by a design example that considers both actuator and measurement system faults. The effect on the robustness of the control solution and the accuracy of fault prediction are discussed
  • Keywords
    discrete time systems; fault diagnosis; frequency-domain analysis; linear quadratic Gaussian control; monitoring; multivariable systems; polynomials; ships; stability; LQG control; discrete time systems; fault detection; fault estimation; fault monitoring; frequency domain; multivariable systems; optimal control; polynomials; robustness; ship roll; stabilization; Costs; Economic forecasting; Estimation error; Fault detection; Fault diagnosis; Frequency domain analysis; Marine vehicles; Monitoring; Polynomials; Robust control;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Decision and Control, 1998. Proceedings of the 37th IEEE Conference on
  • Conference_Location
    Tampa, FL
  • ISSN
    0191-2216
  • Print_ISBN
    0-7803-4394-8
  • Type

    conf

  • DOI
    10.1109/CDC.1998.761765
  • Filename
    761765