压电陶瓷微位移工作台的建模方法

压电陶瓷存在着迟滞非线性特性,而且在运行过程中,特性的变化规律也是未知的、不确定的,难以用传统方法获取既有足够精确性又不至于过分复杂的压电陶瓷微位移工作台的数学模型。基于神经网络,本文提出了一种新的建模方法。分析了微位移工作台的结构和建模方法,利用神经网络的自学习和自适应能力,在线调整模型结构和参数,减小工作台的建模误差,为控制系统提供了更为准确的模型信息。采用工作台的位移数据对网络模型进行了训练,实验结果表明,在80 μm行程范围内,工作台的平均定位误差为80 nm,最大误差为100 nm,基本满足纳米定位的精度要求。

Modeling method of micro-displacement stage driven by piezoelectric ceramic

The piezoelectric ceramics have the property of hysteresis and nonlinearity, and the change rule is unknown and uncertain. It is difficult to build a high accuracy mathematical model simply using the traditional method. In order to improve the model accuracy of micro-displacement stage driven by piezoelectric ceramics, a new modeling method based on neural network is proposed in this paper. The structure and the modeling method of the stage are analyzed. Because of the advantages of the neural network in self-learn and self-adapting, the structure and the parameters can be adjusted on line to reduce the error of the stage model, and more exact information can be provided for the control system. Through training the net model by the stage displacement data, experiment results show that the average error and the maximum error are reduced to 80nm and 100 nm within 80 μm journey respectively, which satisfies the precision requirement of nanometer positioning.