【作者】 李鸣鸣；

【导师】 龚振邦； 孙麟治；

【作者基本信息】 上海大学 ， 机械电子工程， 2007， 博士

【摘要】 随着IC制造工艺、超精密加工、精密仪器及生物医学工程等众多领域的不断发展，与之相关的设备对定位行程、定位精度的要求越来越高，因此大行程纳米定位技术在现代科学技术中起到十分关键的作用，近年来一直是精密工程领域的研究热点之一。本论文对此开展的研究工作是：研制一套宏微(双)驱动方式的大行程纳米定位及其标定的实验研究系统，并对定位技术中的误差建模、全工作行程上定位系统内部的振动测试分析、宏微定位系统的动态建模及特性分析等几个关键问题进行研究。通过对各种大行程纳米定位系统构成方案的深入分析，确定了由交流伺服电机加滚珠丝杠螺母机构为宏动台、压电陶瓷加柔性铰链机构为微动台、纳米光栅测量装置为系统位置全闭环的宏微双驱动方式大行程纳米定位系统；以PC机为上位机，用VB6.0对宏、微定位系统进行了集成以实现宏微定位，定位误差测量及工作台精度评定通过单频激光干涉仪标定系统来实现。定位系统实现了100mm行程、10nm定位分辨率、20mm／s运动速度的工作指标。在定位误差的建模研究中，对定位系统中各种误差的来源、特点进行了详细的分析，以误差是否具有时变性为依据、并根据实际应用中误差补偿的实现方法来研究定位误差的建模问题，建立了插值、多项式、RBF神经网络和灰色系统GM(1，1)等多种静、动态误差模型，并对它们进行了理论分析和仿真研究。在静态误差模型研究中，通过比较得到了改善过拟合RBF神经网络模型这一比较理想的静态误差模型；在对RBF神经网络建模时数据中的噪声处理方面，采用对径向基函数分布宽度和训练目标进行二维优化方法来改善神经网络的过拟合问题，取得了显著的效果。用GM(1，1)建立的灰色动态模型，能将定位系统中的系统误差和随机误差统一考虑，用能量变化方法分析误差的动态特性，所需的建模数据少、计算量小、补偿速度快，是补偿动态定位误差的有效方法。在改善灰色预测模型精度方面，采用了改善背景值的GM(1，1)模型，仿真和实验研究证明它比普通的GM(1，1)模型具有更高的模型精度。对定位精度产生影响的振动问题包括外部的环境振动和定位系统内部的振动。本论文从宏定位系统内部入手，通过构建以LabVIEW软件为核心的虚拟仪器测试系统，用傅立叶变换和小波变换联合分析、对比的方法，研究了在伺服状态下宏动台全行程上定点和运动状态的振动特征。尽管两种方法得出了一致的结论，但小波变换方法可分析工作台各工作位置上的动态特性，分析结果更加具体、准确，其有效性通过PID控制参数调整实验得到了验证。在宏微定位系统的动态特性研究中，根据宏、微工作台的串联形式建立其整体动态模型，分析了宏、微工作台之间的动态耦合关系，讨论了宏微工作台在不同的驱动形式(宏驱动、微驱动和宏微驱动)的阶跃响应。研究结果表明，由于作用位置的不同，宏、微驱动对宏微定位系统动态响应的影响是相反的，应用中要根据实际需要确定宏、微工作台的动态特性参数。根据所建立的动态模型，用宏、微工作台两种不同动态特性参数的配置，分别对白噪声干扰和正弦干扰下的动态补偿定位、低频和高频正弦输入信号的连续补偿定位进行了仿真研究，验证了所建动态模型的正确性和宏微定位实现的可行性，研究结果能够对宏微定位系统设计起到很好的指导作用。最后，用论文中研制的大行程纳米定位及其标定系统进行了实验研究。首先对大行程和微行程内的定位误差进行了标定及分析，而后进行了静、动态误差的补偿实验和宏微两级定位实验，用定位误差评定的国际标准ISO230-2对宏微定位误差进行了处理和分析。实验结果验证了论文研究工作的正确性和本实验系统的有效性。更多还原

【Abstract】 With the progress of techniques in fields of IC manufacturing, ultra-precision machining, precision instrument and Biomedical Engineering, etc., requirement of positioning accuracy related to these apparatus is getting higher more and more. Nowadays, nano-positioning technique with long stroke plays an extremely important role in modern science and technology, and for years, it is the focus in precision engineering research. The studying aims of this paper are to develop an experiment system of long stroke nano-positioning stage based on macro-micro (dual) actuators and its calibration system, and then studying some key problems in positioning technique, such as positioning errors modeling, testing and analyzing to inner vibration of positioning system within full working range of the stage, dynamic system modeling on micro-macro positioning system and discussing of its feature.By comprehensive comparing, the structure of long stroke nano-positioning system and its calibration system is determined. Positioning system is consisted of macro stage, which are made of AC servo motor and ball-screw unit, and micro stage, which are made of PZT and flexible hinge respectively. A grating measurement with 10nm resolution is used for position feedback of entire loop of the system. Macro and micro positioning system (stage) are integrated with a PC as host computer and the VB6.0 as software. Single-frequency laser interferometer is used for errors measurement and accuracy evaluation of the positioning system. The system can realize the working parameters as 100mm working range, 10nm positioning resolution and 20mm/s moving velocity.For the study of errors modeling, reasons for producing positioning errors and its features are analyzed completely. Two kinds of errors, i.e. static and dynamic errors are determined according to the characters of time-varying and compensating method of the errors. Interpolation, curve fit and RBF neural network models are used as static errors model, while grey system model GM(1,1) is used as dynamic errors model. Theory analysis and simulation are carried out to these models. By comparing the static models above, it can be concluded that the improved (de-noised) RBF neural network model is an ideal one relatively. As to the strategies of modeling data de-noising in RBF network, the method of two dimensions optimizing to the parameters of radial basis function spread and network training goal is proved to be especially effective. In the dynamic errors modeling, system errors and random errors are treated as a whole in grey system model GM(1,1). The character of dynamic errors is explained in way of energy change by GM(1,1). The model has the merits of little modeling data, little computing time and fast errors compensating speed. So it is an effective method for dynamic errors compensation. For improving the model’s precision, the method of using reformed background value is taken. It is proved by simulation and experiment that the model is of higher precision.There are two kinds of vibration affect to positioning accuracy. The first is external environment vibration and another is the inner vibration of positioning system. The later one is studied in this paper. Virtual instrument vibration testing system is developed based on the software of LabVIEW. Testing data for some fixed positions and moving path on working range of macro stage in servo state are analyzed and compared by means of Fourier transform and wavelet transform. Though same conclusions are obtained by two of the studying methods, but wavelet transform can analyze the dynamic features of each position of macro stage, and get more detailed, precise results. The effectiveness of this method is validated by the experiment of PID parameter regulating.For the study of dynamic properties of macro-micro positioning system, the integral dynamic model (macro-micro model) of macro stage and micro stage is established according to series relationship of them. Theoretical Study shows that there is a dynamic coupling between macro stage and micro stage, and step response shows that there is an opposite action on macro-micro positioning system with different system input (macro input and micro input) because of different acting position of the input. So, dynamic parameters of macro and micro stages should be arranged carefully based on particular application. Two kinds of arrangement are applied to dynamic model of the macro-micro stage. With the arrangement, simulating study on dynamic compensation positioning disturbed by white noise and sinusoid signal, continuous compensation positioning to the input of sinusoid signal with low and high frequencies. Simulating study proved the rightness of the macro-micro model, the possibility and practicability of macro-micro positioning. The studying results take great role in the design of macro-micro positioning system.In the end, corresponding experiments are carried out with the dual-actuators positioning system developed in this paper. First of all, the positioning errors in long stroke and micro stroke are calibrated and analyzed carefully. Then the experiments of static and dynamic errors compensating are conducted. At last, macro-micro positioning experiment is achieved and the experiment data are evaluated by international standard of ISO230-2. The correctness and effectiveness of the research in this paper are validated by all of these experiments.更多还原