Robust unknown input observer based fault detection for high-order multi-agent systems with disturbances.

Area of Science:

• Control Systems Engineering
• Networked Systems

Background:

• Multi-agent systems (MAS) are increasingly complex, requiring robust fault detection (FD) methods.
• External disturbances pose significant challenges to the reliability of MAS.

Purpose of the Study:

• To develop an effective fault detection strategy for high-order multi-agent systems subjected to unknown disturbances.
• To enhance the resilience and reliability of individual agents within a networked system.

Main Methods:

• Construction of unknown input observers (UIOs) within neighboring agents to detect faults.
• Handling of unknown inputs (UIs) by either perfect decoupling or partitioning strategies.
• Residual analysis to isolate fault effects from system dynamics.

Main Results:

• Demonstrated that perfect unknown input decoupling eliminates disturbance influence on the residual.
• Proposed a novel partitioning method to constrain the impact of unknown inputs when perfect decoupling is not achievable.
• Simulation results validate the proposed fault detection approach's effectiveness.

Conclusions:

• The proposed fault detection method is effective for high-order multi-agent systems with disturbances.
• The UIO-based approach provides a robust solution for identifying agent faults in complex networked environments.
• The partitioning strategy offers a practical way to manage unknown inputs in fault detection.