【作者】 王化明；

【导师】 朱剑英；

【作者基本信息】 南京航空航天大学 ， 机械电子工程， 2005， 博士

【摘要】 智 能 制 造 中 的 微 操 作 / 微 装 配 系 统 , 作 为 微 机 电 系 统 (MEMS, Micro-Electro-Mechanical System)的核心,是微细智能制造的重要实现手段,得到了越来越多的重视和应用。本文在国家自然科学基金重大项目“支持产品创新的先进制造技术中的若干基础研究”(项目批准号:59990470)及自然科学基金项目“智能微机电系统视觉/力觉/位移混合检测与控制技术”(项目批准号:50275078)的资助下,根据调研结果及应用需求建立了应用于微细智能制造的微操作/微装配系统。论文的主要工作及创新点如下: 1.提出了基于切矢转角局部特征的目标识别与定位算法。以自然参数化的边缘曲线上相邻点切矢之间的转角为计算依据,通过对图像与模板的特征曲线进行比较,实现了模板的初步定位。以模板相对于图像的变换参数为变量的目标函数,对目标函数进行最小化,实现了模板的亚像素定位。模板上被遮挡的边缘像素点通过其与图像上最近的边缘像素点之间的距离来确定,由于该点不反映在目标函数中,从而不会影响模板的定位精度。为了满足对三维目标不同深度特征进行观察的需要,在显微视觉系统中增加了局部自动聚焦的功能。 2.采用了 CMAC 神经网络与 PD 控制器结合控制的方法,由 CMAC 神经网络产生前馈输出,PD 控制器产生反馈输出,实现了由图像空间的误差信号向任务空间的控制信号的转换,避免了图像雅可比矩阵的不断调整及其复杂的求逆过程。 3.根据有限元方法对应力、应变所进行的仿真分析,设计并制作了由压电陶瓷堆驱动、基于柔性铰链放大机构的微型夹持器。压电陶瓷堆产生的微小位移由柔性铰链放大机构放大,形成微型夹持器指尖的大范围、高精度的运动。并通过轴孔装配试验对微型夹持器的性能进行了验证。 4.建立了应用于微细智能制造的微操作/微装配系统,并以 Visual C++6.0 和MIL7.5 为开发环境,实现了微操作/微装配系统的控制系统软件。更多还原

【Abstract】 Micromanipulation/Microassembly system for intelligent manufacturing, as the center of MEMS(Micro-Electro-Mechanical System), which is an important way to intelligent micro-manufacturing, has gained more and more attention and application. According to the survey on current available microassembly workstations and application requirements, a micromanipulation/microassembly system for intelligent micro manufacturing is contructed under the support by the key project (No. 59990470) and the project (No. 50275078) of National Nature Science Foundation of China. The main creativities and achievements of the dissertation are as follows: 1.An object recognition and localization algorithm based on local feature―turning angle between neighbouring tangent vectors, is presented. The primary localization of template on image is obtained by comparing the signatures, which are based on the turning angle between the neighbouring tangent vectors at the naturally parameterized edge curve, of image and template. By composing and minizing the object function, which is the function of the transformation parameters of template relative to image, subpixel localization accuracy is achieved. The occluded edge pixel of template is determined according to its distance to the nearest edge pixel of image and will not be taken into account in object function, so the localization accuracy will not be influenced by occlusion. To meet the need of observation of different depth features of 3D object, local autofocusing is added to microscopic vision system. 2.CMAC neural network and PD controller is combined to implement the transformation from the error signal in the image space to the control signal in the task space in order to avoid the iterative adjustment and complicated inverse solution of the image Jacobian. In this control scheme, the CMAC neural network gives the feedforward control and PD controller the feedback conrol. 3.According to the finite element method (FEM) analysis on the stress and strain distribution, a microgripper based on flexure hinge amplification mechanism is designed, which is actuated by piezoelectric ceramic stack. The micro displacement produced by piezoelectric ceramic stack is enlarged by the flexure hinge mechanism to form the large range, high precision motion of the finger. A peg-in-hole experiment is performed to validate the capability of microgripper. 4.A micromanipulation/microassembly system is constructed for intelligent micro manufacturing. Control software is realized with the developing environment Visual C++6.0 and MIL 7.5.更多还原