Reinforcement-Learning-Based Fixed-Time Prescribed Performance Consensus Control for Stochastic Nonlinear MASs with...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Robotics

Background:

• Stochastic nonlinear multi-agent systems (MAS) present challenges in achieving consensus, especially with sensor faults.
• Achieving prescribed performance bounds in finite or fixed time is crucial for practical applications.

Purpose of the Study:

• To develop a fixed-time prescribed performance optimal consensus control strategy for stochastic nonlinear MAS with sensor faults.
• To address the impact of unknown sensor faults using adaptive compensation and reinforcement learning.

Main Methods:

• An improved performance function and coordinate transformation for fixed-time convergence.
• A neural network-based adaptive compensation strategy for sensor fault tolerance.
• A reinforcement-learning-based backstepping method for optimal control design.

Main Results:

• Consensus error converges to prescribed performance bounds in fixed time.
• All closed-loop system signals are bounded in probability, demonstrating stability.
• Simulation results validate the effectiveness of the proposed control method.

Conclusions:

• The proposed method effectively achieves fixed-time optimal consensus control in stochastic nonlinear MAS with sensor faults.
• The integration of adaptive neural networks and reinforcement learning enhances fault tolerance and control performance.