Fault tolerant control deals with the control of a system where faults and system changes may occur. This is obtained in two steps. First, a diagnosis of the system faults or changes are derived. In the second step, the controller is reconfigured or redesigned based on the information from the diagnosis.
Automation and Control researches into the development of methods and algorithms for fault tolerant control which are closely related to applications.
- System and fault diagnosis
- Controller reconfiguration
- Design and control algorithms
System and fault diagnosis
An essential element in fault tolerant control is to be able to diagnose (detection, isolation and identification) of faults and changes in the system. This is a real time task. The diagnosis of faults and system changes are developed based on available input and output signals.
Topics include residual generation, stochastic test methods, decision making, and passive and active fault diagnosis methods. Applications include diagnosis of off-set faults in sensors for wind turbines, detection of parameter changes in ships and detection of performance degradation in process systems
After faults or system changes have been diagnosed, the controller needs to be changed or redesigned, so that it is possible to let the system be operating in a stable mode, but with reduced performance. It is then possible to operate the system in a stable mode for a shorter or longer time depending on the system.
Topics include controller architectures, controller switching, reconfiguration of controllers, on-line controller design and optimization.
The increasing complexity of systems and performance requirements initialise the need for new design methods and control algorithms.
The topics include design methods, control algorithms, controller structures and implementation. Application include e.g. control methods for engines.