Active Learning Vibration Control of Aero-Engine Blade Based on Kriging Surrogate Optimization Strategy

Abstract

In aero-engine failures, a significant proportion is attributed to blade vibration-induced malfunctions. Due to factors such as structural complexity and high stiffness, controlling blade vibration in aero-engines poses a challenging issue in the technical community. The development of efficient blade vibration control technologies is urgently needed. Applying intelligent learning control to engine blade vibration control represents both a novel endeavor in intelligent learning control and a fresh approach to addressing the demands of controlling blade vibration in aero-engines. This paper presents an active learning-based strategy for actively controlling blade vibration in aero-engines. Integrating an active disturbance rejection control controller with a Kriging surrogate model, a framework for active learning control based on input–output data is devised. Through dynamic interaction between the closed-loop controller and the experimental system, optimal control parameters for blade vibration suppression are identified using an active learning strategy. The efficiency of the proposed approach is substantiated through numerical simulations involving an macrofiber composite–actuated blade, as well as through real-time experiments conducted on aero-engine blades. The results show that the optimal control strategy can be found to realize vibration control by training a reasonable amount of data. Furthermore, the robustness of the controller is verified under various disturbances, including measurement noise and broadband excitation. The research results provide an efficient and reliable active learning control method for aero-engine blade vibration control, which is especially suitable for engineering.