关键词: 电静液作动器, 小增益技术, 自适应控制, 运动控制, 神经网络

Abstract: Electro-hydrostatic actuators have the advantages of high integration and high power-to-weight ratio, and are an important develop-ment direction for a new generation of aerospace onboard actuation systems. However, their inherent high-order nonlinearity, model uncertainty, and external disturbances bring challenges to high-precision position tracking control. To address the above problems, this paper proposes an adaptive control method based on the small-gain technique. First, under the backstepping control framework, the neural network canonical representation is used to describe the unmodeled dynamics of the system, and the unknown nonlinear terms are transformed into a norm-constrained form. Second, combined with the small-gain technique, a minimal learning parameter adaptive law is designed, which only needs to estimate the norm-related information of the neural network weights online, avoiding the complex matrix weight updating in traditional adaptive neural network control. Meanwhile, the adaptive law can dynamically adjust the uncertainty compensation strength according to the real-time tracking error of the system. While ensuring the robustness of the system, it alleviates the sensor noise amplification problem that may be caused by high-gain feedback. Then, based on the input-to-state stability principle and the small-gain theory, the uniform ultimate boundedness of all signals in the closed-loop system is proved. Finally, comparative verification is carried out on an EHA experimental platform. The experimental results show that, under different desired trajectory conditions, compared with the traditional proportional-integral controller, the maximum tracking errors of the proposed control method are reduced by 55%, 44%, and 61%, respectively, which verifies the effectiveness of the proposed method in high-precision position tracking control of EHA.

Key words: electro-hydrostatic actuator, small-gain technique, adaptive control, motion control, neural network