摘要
针对柔性直线机电作动器因摩擦导致速度跟踪精度下降的问题,提出了一种基于BP神经网络的摩擦补偿方法。该方法构建了数据驱动摩擦模型,利用BP神经网络描述了速度、位置与摩擦力的非线性映射关系及零速摩擦的动态行为;建立了柔性直线机电作动器的刚柔耦合动力学模型,构建了柔索直线运动实验平台,对比了引入摩擦模型前后的性能。实验结果表明,在50 mm/s 1 Hz的速度正弦指令下,使用摩擦模型后的直线机电作动器速度跟踪精度从0.85 mm提升至0.29 mm,死区时间从0.021 s缩短至0.003 s;在100 mm/s 1 Hz的速度正弦指令下,使用摩擦模型后的直线机电作动器速度跟踪精度从0.93 mm提升至0.25 mm,死区时间从0.012 s缩短至0.003 s。实验结果验证了该摩擦补偿方法可提高柔性直线机电作动器的速度跟踪精度,缩短其死区时间。该方法在武器站系统中应用效果良好,对提升直线机电作动器的快速响应能力具有重要作用。
Aiming at the problem that the velocity tracking accuracy of flexible linear electromechanical actuators decreases due to friction,a friction compensation method based on BP neural network is proposed.This method constructs a data-driven friction model and uses BP neural network to describe the nonlinear mapping relationship between velocity,position and friction force,as well as the dynamic behavior of zero-speed friction.The rigid-flexible coupled dynamic model of the flexible linear electromechanical actuator is established,and the linear motion experimental platform of the flexible cable is constructed.The performance before and after the introduction of the friction model is compared.The experimental results show that under the sinusoidal velocity command of 50 mm/s and 1 Hz,the velocity tracking accuracy of the linear electromechanical actuator after using the friction model is improved from 0.85 mm to 0.29 mm,and the deadband time is shortened from 0.021 s to 0.003 s.Under the sinusoidal velocity command of 100 mm/s and 1 Hz,the velocity tracking accuracy of the linear motor actuator after using the friction model has been improved from 0.93 mm to 0.25 mm,and the deadband time has been shortened from 0.012 s to 0.003 s.The experimental results verify that this friction compensation method can improve the velocity tracking accuracy of the flexible linear electromechanical actuator and shorten its deadband time.This method has a good application effect in the weapon station system and plays an important role in improving the rapid response capability of the linear electromechanical actuator.
作者
黄泓胜
吴继春
谢馨
司喆
HUANG Hong-sheng;WU Ji-chun;XIE Xin;SI Zhe(College of Mechanical Engineering and Mechanics,Xiangtan University,Xiangtan 411105;College of Intelligence Science and Technology,National University of Defense Technology,Changsha 410073)
出处
《制造业自动化》
2025年第12期37-44,共8页
Manufacturing Automation
基金
国家自然科学基金(52305079)。
关键词
柔性直线机电作动器
BP神经网络
摩擦补偿
数据驱动方法
速度跟踪精度
flexible linear electromechanical actuator
BP neural network
friction compensation
data-driven method
velocity tracking accuracy
