【作者】 王娜；

【导师】 浦昭邦；

【作者基本信息】 哈尔滨工业大学 ， 仪器科学与技术， 2008， 硕士

【摘要】 压电陶瓷驱动器以其体积小、重量轻、分辨率高、频响较高的优点成为目前广泛应用的理想微位移驱动元件。但是,它存在着迟滞、蠕变和位移非线性等不足,致使其定位精度不高,稳定的时间长。因此本文将对压电驱动微工作台的控制与校正技术进行研究。本文首先设计一套试验系统,包括压电驱动微位移装置、驱动电源、微位移测量装置和数据采集卡等环节,在对压电陶瓷工作机理和位移特性深入分析的基础上,提出一种基于优化BP(back propagation)算法神经网络结构的PID控制方法。该控制方法是基于闭环控制可以减弱或消除压电驱动三个固有误差的原理,结合神经网络的快速计算能力和自适应能力,实现对压电驱动工作台输出位移的控制与校正。控制算法中包括两个神经网络-神经网络PID控制器(NNC)和神经网络辨识器(NNI)。以三层神经网络自学习PID控制器代替常规PID控制器,实现样本的在线采集和优化;采用BP算法对PID控制器进行在线训练,对辨识器进行离线训练。通过NNC和NNI在线参数调整,起到智能控制作用。本文采用Matlab对两个神经网络结构进行仿真研究,从理论上验证该控制器的可行性,并使用Microsoft Visual C++ 6.0开发平台进行控制算法开发实现闭环控制,对提出的神经网络控制算法和传统PID控制算法分别进行实验研究,通过比较验证本文所提出的神经网络控制方法效果。实验证明,采用基于BP神经网络的PID控制方法实现微位移系统的闭环控制,可有效克服压电驱动工作台所固有的迟滞、蠕变和位移非线性等缺陷,可实现微位移工作台的高精度,快速、稳定控制。更多还原

【Abstract】 The piezoelectric actuator is a perfect micro-positioning drive element and is widely used with its merits of small size, light weight, high precision and high frequency response etc. However the device driven by piezoelectric actuator shows strong hysteresis, creep and nonlinearity so that it has low tiny-orientation precision and long time to reach stabilization.The thesis will have a research on the control and remedy of micro-positioning stage driven by piezoelectric actuator.A suit experiment system is firstly designed including a micro-positioning stage driven by piezoelectric actuator,power for the piezoelectric actuator,tiny-displacement transducer and data acquisition card etc.A study of work principle and traits of the piezoelectric actuator are done on the system.A kind of PID controller base on NN(neural network) of optimized BP (back propagation) arithmetic is designed in the thesis.Based on the principle that closed loop system could weaken or eliminate the three intrinsic errors,the method uses the abilities of calculate and adaption of NN to realize the control and remedy of micro-positioning stage driven by piezoelectric actuator.The control arithmetic includes two NN-NNC and NNI.The conventional PID controller is replaced by the self-study controller of NN of three-layer structure for gathering and optimizing the samples.The arithmetic of BP is used for the training of NNC on line and the training of NNI off line.The thesis could control the stage driven by the piezoelectric actuator intellectively on the base of parameters self-tuning of NNC and NNI on line.The controller is imitated and validated the feasibility of the controller in theory with the help of Matlab.The experiments of NN arithmetic and traditional PID arithmetic are separately tested based on Microsoft Visual C++ 6.0 programme in a closed loop system.The results of experiment indicate that the controller is better than traditional PID controller obviously.The results of experiment testify that adopting PID controller based on NN of optimized BP arithmetic could conquer deficiencies of strong hysteresis,creep and nonlinearity displacement of piezoelectric actuator using the closed loop tiny-displacement system and realize the real-time control stably as well as high precision.更多还原