压电陶瓷执行器的神经网络实时自适应逆控制  

Real-time adaptive inverse control based on neural networks for piezoceramic actuator

作　　者：党选举[1] 

机构地区：[1]桂林电子科技大学智能系统与工业控制研究室,广西桂林541004

出　　处：《光学精密工程》2008年第7期1266-1272,共7页Optics and Precision Engineering

基　　金：国家自然科学基金资助项目(No.60572055);广西区自然科学基金资助项目(桂科自No.0640170)

摘　　要：提出了基于内积的压电陶瓷动态神经网络非线性、非光滑迟滞逆模型,采用反馈误差学习方法,快速地在线得到压电陶瓷的逆模型,避免了通过正模型求取压电陶瓷的Jacobian信息。结合PID反馈控制,在dSPACE系统平台上实现了压电陶瓷的神经网络自适应逆控制。为提高实时性,采用了效率高、速度快的C-MEX S Function编程。实验结果表明:神经网络自适应逆控制的控制精度为0.13μm,而PID控制精度为0.32μm。所提出方法有效地消除了迟滞的影响,控制精度高。In order to improve the actuator precision,a method for eliminating nonlinear and no-smooth hysteresis characteristic of piezoceramic actuator was proposed. An inner product-based dynamic neural network nonlinear and no-smooth hysteresis inverse model for piezoceramic was established, in which the feedback error learning method was used to avoid obtaining Jacobian information of piezoceramic by positive model. On dSPACE system platform, a neural networks adaptive inverse control was realized combined with a PID control. In order to satisfy the requirement of real time control, the program was designed by a high efficiency and fast C-MEX S ftmction. The experimental results indicate that the precision of the proposed adaptive inverse control based on neural networks is 0.13 μm and PID control precision is 0.32 μm. It is shown that the proposed control method can remove effectively the hysteresis characteristic of piezoceramic and has higher control precision.

关 键 词：匪电陶瓷 迟滞特性 神经网络自适应逆控制 dSPACE系统 非光滑迟滞逆模型 

分 类 号：TN384[电子电信—物理电子学] TP273.2[自动化与计算机技术—检测技术与自动化装置]