基於智慧型倒階同動控制之三軸龍門式定位平台

Abstract

本論文以個人電腦控制為基礎，發展具有高精度與高強健性之智慧型同動控制系統於龍門式定位平台。龍門式定位平台係利用三部永磁線型同步馬達組合成H型運動模式之雙軸定位平台，其中由於垂直方向是由兩部平行馬達共同驅動，故同動控制遂成為研究龍門式定位平台之重要課題。有鑑於此，本論文先發展單軸馬達倒階控制系統，再將非整數階微積分計算加入其中，以增加可控制參數自由度之方式改善控制效能。接著，為了加強系統的強健性，使用函數鏈結模糊類神經網路對系統不確定項進行估測與補償。而為了達到雙平行馬達之同動，本論文基於單軸控制之基礎，進一步以Lagrange 方程式建立三自由度龍門動態模型，同時為了避免倒階控制中微分膨脹之問題，於設計過程中引入一階低通濾波器成為動態面控制，而為了提升各軸之控制精準度與雙軸之同動效果，亦引入非整數階微積分系統於動態面控制。最後為了確保系統在參數變化、外在干擾與摩擦力等影響下系統均具備強健性，再利用函數鏈結模糊類神經網路直接補償系統之不確定項，並進行控制系統之穩定性分析。本論文所發展之控制系統皆由個人電腦實現，並由實作結果驗證所設計之控制理論有效性與可行性。 關鍵字：動態面控制、函數鏈結模糊類神經網路、龍門式定位平台、非整數階倒階控制、雙平行線型馬達、同動控制The objective of this thesis is to develop personal computer (PC) based high precision and robust intelligent synchronous control systems for a gantry position stage. The gantry position stage is composed of H-type three permanent magnet linear synchronous motors (PMLSMs) in which the vertical direction is driven by two parallel motors. In this regard, the synchronous control has become an important research task of gantry position stage. In this thesis, a backstepping control (BC) system is developed to control the single axis PMLSM first. Then the non-integer calculus is added to increase the degree of freedom of controlled parameter for the improvement of control performance. Moreover, in order to strengthen the robustness of the system, a functional-link-based fuzzy neural network (FLFNN) is developed to estimate and compensate the system uncertainties. In order to achieve the synchronous control of the parallel motors, a Lagrange’s equation is used to establish the three-degree-of-freedom (3-DOF) dynamic model of gantry stage. Furthermore, to avoid differential expansion problem in the design of the BC, a first-order low-pass filter is introduced to perform dynamic surface control (DSC). For the purpose of enhancing the control accuracy of each axis and the synchronization performance of the parallel axes. A non-integer order calculus is further utilized for DSC. Finally, in order to ensure the robustness of the system under the influences of parameters variations and external interferences. In addition, the FNFS is employed to compensate the system uncertainties directly. All the control systems developed by this thesis were realized by PC to verify the effectiveness and feasibility experimentally.