Jan Maciejowski's Research: Fault-Tolerant Control

Suppose some fault occurs in an aircraft or in a chemical process plant. It suddenly starts to behave in an unanticipated manner. Can the aircraft still be flown, or the plant continue being operated? The existing control systems may be able to cope, but if the change in behaviour is large, then the control system may need to be modified, so that it handles the modified behaviour. This research is concerned with how this can be done automatically, and when it can be done successfully. The research is aimed at handling unanticipated failure modes, for which solutions have not been pre-programmed. For example, an engine failure on a multi-engined aircraft is an anticipated failure, and there are standard procedures for dealing with it. But if the engine failure causes some further fault, such as loss of hydraulic power as in the famous Sioux City incident, then it is unlikely to have been anticipated, and some generic response is required. Another interesting reconfiguration "opportunity" is the DHL Airbus incident.

The motivation for this research is increased safety of operation of aircraft, vehicles, and industrial installations. In particular, the expected increased use of unpiloted aircraft ("UAV's"), in both civil and military roles, makes this research timely. Pilots handle most airborne emergencies successfully; removing the pilot requires this emergency-handling role to be replaced by some autonomous system.

The possibilities for automatic reconfiguration and redesign of control systems are increased by several technological developments:

  • "Clean" measurements from high-quality sensors. These allow anomalies to be detected relatively quickly.
  • Self-diagnosing sensors and actuators, combined with fieldbus networking, allow rapid detection and identification of many faults.
  • Increasing use of detailed "first-principles" models makes more information about the plant available to the diagnostic and redesign algorithms.
  • The continuing increase in computing power and speed allows fancy calculations to be done quickly.
  • Fault-tolerant control involves both automatic system identification and control system redesign. My research relies on model predictive control to achieve the control redesign.


  • MPC Fault-Tolerant Flight Control Case Study: Flight 1862, IFAC Safeprocess Conference, Washington DC, 9-11 June 2003.
  • The Implicit Daisy-Chaining Property of Constrained Predictive Control, Applied Maths and Computing Science, vol.8, no.4, 695-711, 1998.
  • Modelling and Predictive Control: Enabling Technologies for Reconfiguration, IFAC Conference on Systems Structure and Control, Bucharest, 23-25 October 1997. (Reprinted in Annual Reviews in Control, Pergamon, 1999.)
  • Reconfigurable control using constrained optimization, European Control Conference, Brussels, 1-4 July 1997.

  • Back to Jan Maciejowski's Home Page

    Back to the CUED Control Group Home Page

    This page last updated on 30 November 2004.