ADP-Based Fault-Tolerant Control for Multiagent Systems With Semi-Markovian Jump Parameters

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Stochastic Processes

Background:

• Multiagent systems (MAS) face challenges in achieving consensus due to environmental variations and component faults.
• Semi-Markovian processes offer a flexible framework for modeling systems with time-varying parameters and complex state transitions.
• Actuator bias faults can degrade system performance and hinder consensus achievement in MAS.

Purpose of the Study:

• To develop an integrated adaptive dynamic programming (ADP) and adaptive fault-tolerant control (FTC) strategy for semi-Markovian jump multiagent systems.
• To address the consensus control problem in the presence of actuator bias faults and environmental uncertainties.
• To overcome the need for precise system dynamics knowledge through a data-driven approach.

Main Methods:

• Integration of adaptive dynamic programming (ADP) with an actor-critic neural network structure to learn system dynamics online.
• Implementation of a data-driven adaptive fault-tolerant control (FTC) scheme for real-time compensation of actuator bias faults.
• Utilizing semi-Markovian processes to model the system's environmental parameter variations and stochastic behavior.

Main Results:

• Demonstrated uniform boundedness of the controlled system's signals, ensuring stability.
• Achieved consensus, where follower agents' states successfully track the leader's state.
• Validated the proposed control strategy's effectiveness through simulation results.

Conclusions:

• The integrated ADP and FTC approach effectively solves the consensus control problem for semi-Markovian jump multiagent systems with actuator bias faults.
• The data-driven nature of the FTC scheme allows for online fault compensation without prior system identification.
• The study confirms the robustness and efficacy of the proposed method in complex, uncertain environments.