Differential Evolution Algorithm for Fast Gains Learning in a High-Gain Controller.

Area of Science:

• Control Systems Engineering
• Machine Learning
• Robotics

Background:

• Traditional algorithms like Twin Delayed Deep Deterministic Policy Gradient (TD3) and Genetic (G) algorithms often exhibit slow convergence rates.
• Fast convergence is crucial for efficient trajectory tracking in high-gain control systems.
• Existing methods may not meet the demands for rapid adaptation and learning in dynamic control applications.

Purpose of the Study:

• To propose a novel Differential Evolution High-Gain Controller (DEHGC) for accelerated learning and trajectory tracking.
• To investigate the efficacy of the Differential Evolution (DE) algorithm for fast gains learning in high-gain controllers.
• To compare the convergence speed and performance of the DEHGC against TD3 and G algorithms.

Main Methods:

• Implementation of a high-gain controller integrated with a Differential Evolution (DE) algorithm for parameter tuning.
• Detailed pseudocode provided for the proposed DEHGC algorithm.
• Comparative analysis of DE, TD3, and G algorithms in the context of fast gains learning for high-gain controllers.

Main Results:

• The Differential Evolution (DE) algorithm demonstrates faster convergence compared to TD3 and G algorithms.
• The proposed DEHGC ensures error stability within the high-gain controller.
• The DEHGC facilitates rapid learning of controller gains for enhanced trajectory tracking.

Conclusions:

• The DEHGC offers a promising alternative for fast gains learning in high-gain controllers.
• The DE algorithm's faster convergence is beneficial for achieving rapid trajectory tracking.
• The DEHGC approach enhances control system performance through accelerated learning and stable error correction.