LHS-GA Based H-Infinity Control for Robust Airfoil Flutter Suppression

Abstract

Presented is a controller design methodology for practical ASAF (Active Suppression of Airfoil Flutter) applications. The approach utilizes a heuristic evolutionary optimization Genetic Algorithm (GA) combined with Latin Hypercube Sampling (LHS) to synthesize a gain scheduling controller for a four-degree-of-freedom aeroelastic airfoil model, accounting for uncertainties and nonlinearities. The objective is to design a robust controller that effectively suppresses flutter while optimizing performance metrics, specifically time-domain constraints and settling time. The LHS-GA method bridges the gap between traditional frequency-domain approaches and computationally expensive online optimization methods. Unlike frequency-domain methods that rely on trial-and-error or conservative assumptions, LHS-GA automatically synthesizes a robust static controller by directly addressing time-domain constraints and performance metrics. It eliminates the need for online optimization, ensuring robust stability under parametric uncertainty. This novel approach provides a systematic and efficient solution for designing controllers, offering an alternative to online optimization control methods and offline conservative frequency domain methods. By employing LHS, the proposed method efficiently explores the uncertain parameter space, promoting robustness in the solutions. The heuristic evolutionary optimization method selects optimal controllers by mapping the weight functions of the $H_{\infty } $ synthesis to time-domain metrics. Two case studies with different time-domain constraints are presented, demonstrating the approach capability to handle uncertainty and synthesize a robust controller that meets time-domain constraints and optimizes performance metrics. Finally, the results are compared against a traditional LQR controller design and a conventional $H_{\infty } $ controller design highlighting the advantages of the proposed method.

Keywords

• Genetic algorithm
• gain scheduling
• H-infinity
• Latin hypecube sampling
• time-domain constraints