| 摘要: | 本論文之目標為發展以個人電腦為控制核心之智慧型奈米精密定位控制技術之研製，設計出適用於半導體工程、掃描探針顯微鏡及微機電系統的智慧型精密定位之控制技術。
在本論文中，首先針對壓電效應作詳細說明，並介紹壓電元件的種類與壓電奈米定位平台之優缺點。接著介紹近年來相關文獻中各種壓電奈米定位平台之磁滯模型。此外，提出磁滯摩擦力模型來描述壓電奈米定位平台之磁滯現象，並利用LuGre磁滯摩擦力模型與機械動態模型來建立壓電奈米定位平台之動態模型。由於磁滯摩擦力模型有許多參數於實際應用無法獲得且壓電奈米定位平台動態模型亦存在許多不確定項，因此，提出了遞迴式類神經網路(Recurrent Neural Network, RNN)之智慧型積分步階迴歸滑動控制(Intelligent Integral Backstepping Sliding-Mode Control, IIBSMC)來估測包括系統參數和外部干擾之線上總集不確定量，藉以提高壓電奈米定位平台的控制性能和強健性。接著，利用計算力控制技巧發展非對稱型函數連結放射狀基底函數網路計算力控制系統(Functional Link Radial Basis Function Network with an Asymmetric Membership Function, FLRBFN-AMF) 用來估測非線性函數包含壓電奈米位移平台的動態總集不確定量以提高追隨多樣的參考軌跡之性能。另外，使用Elman類神經網路(Elman Neural Network, ENN)之智慧型非奇異點終端滑動模式控制(Intelligent Nonsingular Terminal Sliding-Mode Control, INTSMC)系統，採用多輸入多輸出(Multi-Input-Multi-Output, MIMO)之ENN來估測以增加壓電奈米定位平台之控制性及強健性。Elman類神經網路估測器是用來估測線上壓電奈米定位平台的磁滯現象及包含系統參數和外部干擾等等的總集不確定量範圍。
由於上述之智慧型控制系統皆利用類神經網路理論近似壓電致動器動態模型之非線性函數，且線上調整智慧型控制系統之參數，故使得系統在磁滯現象、參數變化和外來干擾等因素下仍然具有相當良好的強健性。本論文最後以實測驗證上述所提出之智慧型控制系統的有效性。
;The objective of this dissertation is to develop a high performance intelligent control system for the PC-based piezo-flexural nanopositioning stage (PFNS) that has widely utilized in the industries for many applications such as high-density semiconductors, scanning probe microscopy, and micro electromechanical systems (MEMS), etc.
In this dissertation, first, we examined and explained in detail of the piezoelectric effect, different types of piezoelectric components, and the advantages and disadvantages of PFNS. Then, we introduced hysteresis models for various PFNS mentioned in the relevant literature in recent years. Furthermore, we proposed a hysteresis friction model to describe the hysteresis effect in the PFNS. In order to establish a dynamic model for the PFNS, we adopted LuGre hysteresis friction model and Mechanical model to derive the dynamic model of the PFNS. Due to the hysteresis friction model parameters and dynamic model of the PFNS are unknown, we proposed an intelligent integral backstepping sliding-mode control (IIBSMC) system, where an recurrent neural network (RNN) estimator is proposed to improve the control performance and robustness of the PFNS. In this approach, an RNN estimator is proposed to estimate online the lumped uncertainty, including the system parameters and external disturbances. Moreover, we developed the computed force control system using a functional link radial basis function network with an asymmetric membership function (FLRBFN-AMF) controller to estimate the lumped uncertainty including the parameter variations and external disturbance of the system directly. Furthermore, an intelligent nonsingular terminal sliding-mode control (INTSMC) system using a multi-input-multi-output (MIMO) Elman neural network (ENN) estimator is proposed to improve the control performance and robustness of the PFNS. The ENN estimator is proposed to estimate the hysteresis phenomenon and lumped uncertainty including the system parameters and external disturbance of the PFNS online. In addition, the parameters of the intelligent control systems are learning on line. Using the proposed control systems, the robustness to hysteresis, parameter variations and external disturbances can be obtained. Finally, the effectiveness of the proposed control schemes are demonstrated by some experimental results. |
