基于分数阶自适应神经网络的电动舵机伺服系统摩擦干扰补偿控制

A Friction Disturbance Compensation Method for Electromechanical Actuator Based on Fractional Order Adaptive Neural Network

摩擦干扰力矩影响电动舵机伺服系统的跟踪性能,造成位置和速度跟踪偏差,甚至可能导致伺服系统不稳定。针对摩擦力矩干扰下的电动舵机伺服系统跟踪性能差的问题,本文提出了一种分数阶自适应神经网络摩擦补偿算法(FOANN),估计并补偿摩擦干扰力矩。首先,建立基于LuGre模型的电动舵机伺服系统模型,利用径向基神经网络估计模型中的不可测状态变量。其次,设计FOANN摩擦补偿控制器,利用李雅普诺夫稳定性理论证明电动舵机闭环系统的稳定性。最后,利用仿真和实验平台,对比分析了FOANN、传统PD控制和模型自适应控制的性能。结果表明,基于本文所提出的FOANN摩擦力矩补偿控制算法,电动舵机伺服系统的位置跟踪误差和速度跟踪误差均大幅减小,FOANN算法能够有效估计并补偿摩擦力矩,降低摩擦干扰对电机舵机伺服系统的影响,提高伺服系统的动态性能。

Friction torque disturbance affects the tracking performance of electromechanical actuator servo system,bringing position and speed tracking errors,and even may leading to instability of the servo system.Aiming at the problem of poor tracking performance of electromechanical actuator servo system under friction torque disturbance,a FOANN friction compensation algorithm is proposed to estimate and compensate the friction torque.Firstly,base on LuGre friction model,a electromechanical actuator model is established,and the unmeasured state variable in the LuGre model is estimated by radial basis function neural network.Secondly,a FOANN controller is designed,and the stability of corresponding closed-loop system is proved by Lyapunov stability theory.Finally,through simulation and experimental platform,the dynamic performance of FOANN is compared with those of traditional PD and MRAC.The simulation and experimental results show that,with the proposed FOANN friction torque compensation algorithm,the tracking errors of both position and velocity of electromechanical actuator servo system are greatly reduced.FOANN algorithm can effectively estimate and compensate friction torque,reduce the impact of friction disturbance and enhance the dynamic performance of the servo system.