【作者】 杨雪锋；

【导师】 李威；

【作者基本信息】 中国矿业大学 ， 机械设计及理论， 2009， 博士

【摘要】 随着现代生物医学工程的发展,许多细胞级的操作如细胞分离、捕获、切割及注射等均需通过显微操作装置完成。微操作台作为细胞微夹持器、微注射器及培养皿等微操作设备的承载机构,其位移输出范围、位移分辨率及运动精度是影响细胞操作成功率的重要因素。针对生物细胞显微操作的特点及要求,微操作台需具有毫米级的运动范围和微米甚至亚微米级的定位及重复定位精度。因此,采用结构嵌套方式,将压电陶瓷/压电驱动器作为操作台的驱动部件,结合放大机构和平行导向机构构建了柔性一体化大行程二维微操作台,建立了微操作台的动态迟滞模型并对其控制特性可控制方法进行了实验研究。实验结果表明,设计开发的一体化大行程二维微操作台具有毫米级的运动范围和亚微米级的分辨率,可以满足细胞操作的要求。压电驱动器作为微操作台的驱动部件,其位移分辨率和控制精度是影响微操作台的位移输出特性的重要因素。因此,本文首先从压电驱动器的负载特性出发,分析了容性负载对驱动电源的影响,得出驱动器的等效负载电容可导致驱动系统产生相位滞后甚至振荡,为此设计了由高压放大器组成的桥式驱动电路,并针对压电驱动器这种特殊负载,采用滞后-超前控制器对驱动电源的频率特性进行了校正,使系统获得了良好的静态及动态特性。由于压电驱动器的输出位移不能满足细胞操作的要求,在分析不同柔性铰链位移输出特性和力学特性的基础上,选用直角平板柔性铰链作为机构的传动部分,研究了其几何参数对桥式放大机构和U型导向机构的运动特性及力学特性的影响,然后采用结构嵌套方式,构建了由桥式放大机构和U型导向机构组成的大行程、无耦合二维微操作台,分析了微操作台的位移输出特性及应力集中特性,建立了微操作台的动力学模型,并通过有限元分析和实验验证了理论分析的正确性。在细胞操作过程中,根据操作的需求,微操作台一般做随机运动,因此施加在压电驱动器上的幅值和电压变化率也不相同,但不同电压变化率会导致微操作台的迟滞特性发生变化。目前流行的基于Preisach模型的静态建模方法并不能描述这种动态迟滞特性,因此引入神经网络辨识理论,利用Preisach模型中提出的转折点、上升和下降过程等迟滞多环曲线的形态特征,结合神经网络构造出一种对驱动电压变化率敏感的动态迟滞模型,然后对模型的泛化能力进行了验证。在微操作台的迟滞模型研究的基础上,首先研究了微操作台实际控制特性,然后采用神经网络动态迟滞逆模型、传统PID以及神经网络自适应PID等控制方法对微操作台的在不同控制信号下的输出位移进行了分析,并通过实验比较了不同控制方法对随机位移信号的跟踪特性。本论文有图125幅,表7个,参考文献153篇。更多还原

【Abstract】 With the development of the modern biomedical engineering, many cellular level manipulations such as cells isolating, capturing, microdissecting and microinjecting must be done by micromanipulator. The micro-stage which as the supporting table of microgripper, microinjector and culture dish and its displacement range, resolution and kinematic accuracy are the important factors effect the success of cell-manipulation.According to the characteristics and requirements of cell-manipulation, the micro-stage should have the characteristic of large stroke to millimeter and the displacement resolution to micron or submicron. Therefore, a large stroke two-dimension integrated micro-stage was established by employing piezo actuator as the driving component and the nested structure of amplification and parallel guiding mechanism. Then the dynamic hysteresis model of micro-stage was established and the experiments about the control strategy were performed. The experimental results proved that the large stroke two-dimension integrated micro-stage has the displacement of millimeter and the resolution reached to submicron, those features could meet the requirements of cell-manipulation.The displacement resolution and kinematic accuracy of piezo actuator are the most important factors which influence the output displacement performance of micro-stage. Then this dissertation analyzed the effect of capacitive load to driving power supply based on the load characteristic of piezo actuator and derived that the equivalent capacitance of piezo actuator can lead to phase lag even oscillation of the driving system. For this, a bridge type driving circuit was designed based on high voltage operational amplifier. Meanwhile, a lead lag controller was applied to corrected the frequency characteristic of the capacitive load system and obtain the better static and dynamic performance.Because the output displacement of piezo actuator could not meet the requirements of cell-manipulation, the right angle flexure hinge was adopted as transmission section based on the analysis of displacement and mechanical performance to various type flexure hinges. The effect of geometric parameters to kinetic and mechanical performance of bridge type amplification and U-type parallel guiding mechanism were studied. Then a large stroke two-dimension integrated micro-stage was established based on bridge type amplification and U-type parallel guiding mechanism with the nested structure. The output displacement and stress concentration characteristics were analyzed. Finally, the dynamic model of micro-stage was established and verified by the finite element method and experiment.of cell-manipulation because of the random motion of micro-stage according to the operation requirements. However, the different voltage regulation can lead to the change of hysteresis characteristic of the micro-stage. And the static modeling method based on Preisach-type model did not describe this dynamic hysteresis phenomenon. So the neural network identification theory was introduced into and a dynamic hysteresis model was established based on the morphological of turning point, rising and falling of multi-loop hysteresis curve proposed by Preisach model. Then, the generalization ability of the dynamic model was verified by experiments.The control performance of micro-stage was tested by experiments base on its dynamic hysteresis characteristic. Then the control performance for step response and sine response was studied by employing the dynamic hysteresis inverse model, classical PID and the neural network adaptive PID control strategy. Finally, the random displacement tracking performance was compared by employing different control method.There are 125 figures, 7 tables and 153 references in the dissertation.更多还原