A switched extremum seeking control without steady-state oscillation using gradient-based adaptive-amplitude...

Area of Science:

• Control Engineering
• Optimization Theory
• Nonlinear Systems

Background:

• Extremum seeking control (ESC) is crucial for optimizing system performance.
• Existing ESC methods face challenges with nonconvex problems and gradient estimation accuracy.
• Improving both dynamic and steady-state performance of ESC remains an active research area.

Purpose of the Study:

• To propose a novel switched extremum seeking control (ESC) scheme.
• To enhance both steady-state and dynamic performance of ESC.
• To address challenges in nonconvex optimization and gradient estimation.

Main Methods:

• Development of a novel adaptive law for perturbation using hybrid gradient information.
• Implementation of a Lyapunov-based switched strategy to initiate bisection-based ESC.
• Theoretical stability analysis of the proposed switched ESC scheme.

Main Results:

• The adaptive-amplitude ESC achieves global optimization, overcoming local extrema in nonconvex problems.
• The switched strategy ensures rapid and accurate plant output convergence without steady-state oscillation.
• Theoretical proofs confirm the stability of the proposed ESC scheme.

Conclusions:

• The proposed switched ESC scheme effectively enhances steady-state and dynamic performance.
• The method provides a robust solution for nonconvex optimization problems.
• Numerical simulations validate the effectiveness and stability of the advanced ESC approach.