Periodic event-triggered adaptive tracking control design for nonlinear discrete-time systems via reinforcement...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Nonlinear Dynamics

Background:

• Traditional control systems often require continuous data transmission, leading to inefficient resource utilization.
• Event-triggered control (ETC) offers a more efficient alternative by transmitting data only when specific conditions are met.
• Reinforcement learning (RL) provides a powerful framework for developing adaptive control strategies.

Purpose of the Study:

• To develop a novel periodic event-triggered control (ETC) scheme for nonlinear discrete-time systems.
• To integrate an actor-critic architecture within reinforcement learning (RL) for adaptive control.
• To enhance resource efficiency in control systems through intelligent data transmission.

Main Methods:

• A periodic event-triggered mechanism (ETM) was designed to regulate data transmission to controllers.
• An actor-critic neural network architecture, utilizing radial basis function neural networks (RBFNNs), was employed.
• Lyapunov stability analysis was used to rigorously prove system stability and boundedness of error signals.

Main Results:

• The proposed periodic ETM guarantees a minimum inter-event interval, optimizing communication resource usage.
• The RL-based actor-critic approach successfully approximates the optimal control policy and performance index.
• Stability analysis confirmed that all closed-loop system error signals are uniformly ultimately bounded (UUB).

Conclusions:

• The developed event-triggered control scheme effectively stabilizes nonlinear discrete-time systems.
• The integration of periodic ETM with RL offers significant advantages in communication resource economy.
• Simulation examples validate the practical effectiveness and performance of the proposed control strategy.