Neural-Network-Based Robust Optimal Tracking Control for MIMO Discrete-Time Systems With Unknown Uncertainty Using...

Area of Science:

• Control Systems Engineering
• Nonlinear System Dynamics
• Artificial Intelligence in Control

Background:

• Nonlinear multi-input multi-output (MIMO) discrete-time systems present challenges in robust optimal tracking control due to inherent uncertainties.
• Existing control strategies often struggle to balance performance optimization with stability guarantees under unknown uncertainties.

Purpose of the Study:

• To develop a robust optimal tracking control strategy for nonlinear MIMO discrete-time systems with unknown uncertainty.
• To establish an adaptive actor-critic control method that minimizes a novel cost function and ensures closed-loop system stability.

Main Methods:

• Integration of an adaptive critic design (ACD) scheme into the control architecture.
• Utilization of neural network approximators, specifically an action network for control signal generation and a critic network for cost function approximation.
• Proposal of a non-quadratic cost function to reduce overall design costs.

Main Results:

• The proposed adaptive actor-critic method effectively handles unknown uncertainties in nonlinear MIMO discrete-time systems.
• Optimal control signals and tracking errors are proven to be uniformly ultimately bounded, even in the presence of uncertainties.
• Numerical simulations validate the effectiveness and robustness of the developed control approach.

Conclusions:

• The adaptive critic design-based robust optimal tracking control strategy offers a stable and efficient solution for nonlinear MIMO discrete-time systems.
• The novel non-quadratic cost function contributes to reduced design costs while maintaining performance.
• This approach provides a significant advancement in handling uncertainties for complex control systems.