Multidimensional-Taylor-network-based robust optimal tracking control for MIMO nonlinear discrete-time systems.

Area of Science:

• Control Systems Engineering
• Nonlinear System Dynamics
• Adaptive Control Theory

Background:

• Practical engineering systems often suffer from unsatisfactory control due to model uncertainty, speed unavailability, input nonlinearity, and multi-input coupling.
• Existing control strategies struggle to effectively address these complex challenges in discrete-time systems.

Purpose of the Study:

• To propose a robust optimal tracking control strategy for nonlinear multi-input-multi-output discrete-time systems with unknown uncertainties.
• To minimize the cost function and stabilize the closed-loop system by developing an adaptive control approach.

Main Methods:

• Utilized a Multi-dimensional Taylor Network (MTN) comprising an actor MTN and a critic MTN for adaptive control.
• Employed the approximation property of MTN to generate optimal control signals via the action MTN.
• Approximated the cost function using the critic MTN, tuned online due to the inaccessibility of the exact cost function.

Main Results:

• Developed a novel cost function to reduce computational load during control processes.
• Integrated adaptive critic design principles to effectively handle system uncertainties.
• Simulation results demonstrated the significant effectiveness of the proposed robust optimal tracking control strategy.

Conclusions:

• The proposed robust optimal tracking control strategy successfully addresses uncertainties and nonlinearities in discrete-time systems.
• The adaptive approach using Multi-dimensional Taylor Networks offers a promising solution for complex system control challenges.
• The strategy provides a stable and effective method for achieving optimal tracking performance in practical engineering applications.