摘要
为了提高具有表面微结构零件的超精密加工中快刀伺服器的轨迹预测和跟踪精度与抗干扰性,设计一种新型复合控制:在前馈控制器中用Preisach逆模型补偿系统中压电陶瓷驱动器引起的非线性特性;针对前馈控制器未能补偿的非建模扰动、模型参数的不确定性以及其他外界未知扰动,设计了自抗扰控制作反馈控制器。推导了Preisach逆模型;用RBF神经网络实现了Preisach逆模型对压电陶瓷驱动器的线性化补偿;通过对快刀伺服器的建模分析,得到惯性环节和二阶振荡环节串联的等效模型,设计三阶四维扩张状态观测器,可对未知扰动的观测结果作出实时估计和补偿。根据快刀伺服器和超精密车削的特点,取消跟踪微分器,增加速度输入和加速度输入,设计了改进的自抗扰控制器。上述两种控制器组合成复合控制器。实验表明,该复合控制方法可以提高预测和跟踪精度与抗干扰性。
In order to improve trajectory prediction and tracking accuracy and robustness of fast tool servo (FTS) system, in ultra-precision machining for the components with micro-structured surface, a new composite control approach was introduced. Inverse Preisach model (IPM) was used to compensate hysteresis of the piezoelectric ceramic actuator, based on which the feedforward controller was designed. The active disturbance rejection control (ADRC) law was designed for the not compensated nonlinearity, the uncertainty of the model parameters, and other unknown disturbances. A detailed derivation of the IPM was given ; the linear compensation for the piezoelectric ceramic actuator was realized by IPM, using radial basis function (RBF) neural networks; then by modeling and analysis for the FTS, the equivalent model of inertia linkand two-order oscillation link in series was obtained; a three-order four- dimensional extended state observer (ESO) was designed. It could be used to make real-time observations to estimate and compensate for unknown disturbances from all kinds of sources. According to the characters of FTS and ultra-precision machining, the tracking differentiator was altered with speed input and acceleration input, the ADRC controller was designed. The above two controllers were combined into a composite controller. Experimental results show it can improve the predicting and tracking performance and robustness.
出处
《机床与液压》
北大核心
2017年第10期153-157,161,共6页
Machine Tool & Hydraulics
基金
广东省引进创新科研团队计划(201001G0104781202)
广东高校省级重点平台和重大科研项目(2015KQNCX104)
广东金融学院科研课题(15XJ02-04)
关键词
压电陶瓷
Preisach逆模型
RBF神经网络
自抗扰控制
快刀伺服
Piezoelectric ceramics
Inverse Preisach model (IPM)
Radial basis function (RBF) neural networks
Active disturbance rejection control (ADRC)
Fast tool servo (FYS)