【作者】 于保军；

【导师】 杨志刚；

【作者基本信息】 吉林大学 ， 机械设计及理论， 2008， 博士

【摘要】 由于MEMS(微机电系统)、生物医学工程、精密光学工程和超精密加工等领域的快速发展,迫切需要能够实现微米级、甚至纳米级精密定位系统,各种大行程精密定位系统应运而生。上世纪80年代的中后期,国内外学者相继开展了宏/微双驱动微操作系统的开发研究。经过近20年的探索,不论是从理论还是从应用,都证明了宏/微双驱动微操作系统在很多方面的性能优于传统的采用单一驱动方式的微操作系统。本文在阅读大量的相关文献资料的基础上,致力于基于机器视觉的宏/微双驱动微操作系统、序列图像清晰度评价、图像融合、显微视觉伺服控制等的研究。利用混合电机和丝杠机构的行程长、驱动力大等特点来构成微操作系统的宏动部分;而利用压电陶瓷(PZT)的响应速度快、精度高、无摩擦等特点制作的压电驱动器来构成微动部分。微操作系统既实现了大行程,又达到很高的定位精度。针对目前很多微操作系统的协调控制难的缺点,提出了利用机器视觉来协调宏动和微动的各自控制部分。将微操作系统的末端执行工具与目标部件的距离作为控制阈值,大于阈值时启动宏动驱动器进行驱动和控制;小于阈值时启动压电驱动器进行驱动和控制。本文以生物工程中的细胞注射操作为应用背景,设计了机器视觉系统,并对各项关键技术进行了系统深入的研究。由于将宏/精微操作系统与采用基于图像模式的视觉伺服控制系统相结合,提高了整体微操作系统的移动速度、距离和控制精度,解决了实际工程中遇到的移动位移小、定位精度低以及控制难等问题。最后作为此宏/微双驱动微操作系统的应用,以模拟生物工程中的细胞注射操作为例,对微操作系统的硬件和软件系统进行整体的系统测试。结果证明本系统的识别、跟踪、定位及控制都达到了预期的目的。本文由国家高技术研究发展计划项目(863计划)《多自由度压电精密驱动器》(项目编号:2002AA423150)、国家自然科学基金重点项目《压电精密致动技术基础研究》(项目编号:50735002)和长春市新星创业基金《压电步进驱动器的开发及应用》(项目编号:04-08XC279)提供资助,是这三个项目研究工作的一部分。更多还原

【Abstract】 Recent years,many micro manipulation system or micro manipulation hand apply to military affairs,optics communication,biology and precision mechanism,and these also have success.So these application field also advance more requirement to micro-manipulation system,for example:motion precision,respond speed,force sense,control and flexible.Some project application field not only require large workspace,but also ultra-precision.Some others projects require micro-manipulation system has fast motion speed and precision force sense.All these requirements make the traditional micro-manipulation system with single drive method can not accomplish the appointed task.At the same time microscopic vision is fundamental important to Micromanipulation which provide lower level and non-contact feedback information to control system,include microscale geometry,spatial relations and motion of micro-components.Because of particular properties of microscopic vision,which often suffer from small field of vision and limitied depth of focus.It is important to study that How to conquer the disadvantage from above.In order to resolve above problems,at the end of the twenty century eighty,many scholars of the word put forward macro-micro dual drive micro-manipulation system.In the past twenty years,not only theory prove but also project application both prove that macro-micro dual drive micro manipulation system has many merits over the traditonal single drive micro-manipualtion system.,it is a valid method to resolve large workspace and ultra-precision.The paper mainly research the macro-micro dual drive micro manipulation system based on mechine vision.It adopts motor and silk-bar framework to fabricate the macrostage with large workspace and large drive force etc;its microstage is made of PZT,because PZT has many merits,for example fast response,high precision and frictionless etc characters.So the micro manipulation system at the same time has large workspace and nano precision.In the microscopic feedback progress of micromanipulation,we can acquire in focus image and our of focus image in the same time.It can make some area of target image is clear and the others is blur.We adopt the microscopic auto focus technology base on image definition.A image definition method was developed,which combine the image definition base on traditional grads and the image definition base on wavelet transformation.This technology can provides high response speed,significantly better depth resolution and accuracy.We adobe the target area select method to carry out autofocus for avoid system focus on non-target area.For acquire a clear image of entire target,must used different method of sequence image fusion for different type target.A image fusion method base on wavelet transform was adobed for the target that can be aquire the shape and character.this may generate some distortion,but can not influence the result of judgement of system operate progress,this method can get a good control effect to the micromanipulation system.The other image fusion method base on the same coordinate pixel of sequence image definition evaluate was adobed for the target that can not permit distortion(for example:Cell observe).This method can not increase calculation distortion compare with the clear origin image in the sequence image,that can ensure system work normally.According to the different character of target adobe different tracking arithmetic to resolve the problem of the damage to the system performance of tracking if digital image process result in time delay.Used the area pattern matching fast tracking arithmetic to tracking the target which cannot change the shape(for example:operate-tool,clamp-tool, mechanism part).For the target that maybe change the shape and character in the tracking progress,cannot get the pattern of target beforehand,used the digital matting technology to recognise contour of target and provide the coordinate message of target to the control and decision-making system.The experiment results showed that arithmetics not only improve the system response speed but also ensure the accuracy of tracking.At the same time,we aim at many micro manipulaiton system has trouble in harmonizing control,put forward adopting mahine vision to harmonize each control part of macro and micro.The system looks at the distance between the implement and target as control threshold,if the threshold more than the threshold,the control part start up the macrostage and macro control part,or else,start up the microstage and PZT control part in order to compensate the displacement error.At last,we make use of LabVIEW to design the vision servo programe realizing track,recognising and control.At last,the paper has the background of cell injection of biology project,and design machine vision track and recogniseing programe,and home research those key technology.The system have realized dynamicly recognising injection needle and cell,and truly control microstage motion to appointed place;at the same time,the system has dyanmic place and the distance between needle and cell saving to the appointed database.Experiment result shows the micro manipulation system working stabilization,track,recognising, positioning and control all have achieved anticipative aim.更多还原