Reconfigurable control system design for fault diagnosis and accommodation

Area of Science:

• Control Engineering
• System Dynamics
• Computational Intelligence

Background:

• Increasing demand for system reliability and survivability necessitates advanced fault diagnosis and accommodation techniques.
• Existing fault tolerant control methods often rely on specific assumptions about system dynamics or failure types, limiting their applicability.

Purpose of the Study:

• To investigate the on-line fault tolerant control problem for dynamic systems under unanticipated failures without prior assumptions.
• To develop a robust control strategy capable of handling unknown and catastrophic component failures.
• To ensure system stability and trajectory tracking in real-time environments.

Main Methods:

• Derivation of sufficient conditions for on-line stability using discrete-time Lyapunov stability theory.
• Development of an on-line fault accommodation control strategy integrating control theory and computational intelligence.
• Utilization of an on-line estimator to compute control signals with partially available fault information.

Main Results:

• Theoretical analysis confirms that control objectives can be met on-line without full knowledge of failure dynamics.
• On-line simulations demonstrate the effectiveness of the proposed fault accommodation strategy.
• A real-time experimental testbed validated the technique's feasibility for unanticipated fault accommodation.

Conclusions:

• The proposed on-line fault tolerant control strategy effectively accommodates unanticipated failures in dynamic systems.
• Real-time experiments confirm the promising potential of this approach for systems with limited information on dynamics and failures.