標題: | 機器人控制與錯誤診斷系統 Control and Actuator Fault Diagnosis of Robot Manipulators |
作者: | 翁茂樵 Weng, Mao-Chiao 蕭得聖 Hsiao, Te-Sheng 電控工程研究所 |
關鍵字: | 機械手臂;控制器;錯誤診斷;Robot Manipulators;Control;Fault Diagnosis |
公開日期: | 2008 |
摘要: | 本研究在一個二軸平面的機械手臂實驗平台上,進行控制器與錯誤診斷的整合測試。在控制器的部分,使用非追蹤型卡曼濾波器(UKF)進行角速度估測,將此估測與透過差分所得到的角速度,分別代入計算扭力控制器演算法觀察其效果差異。此外將提出一個僅需位置回授的新型控制器與上述兩個控制器一起比較。在錯誤偵測與判別中,實現二階廣義偽貝氏演算法(GPB-2)及交互多模型(IMM)錯誤演算法的測試,其中所使用之估測器為控制器部分的UKF估測器。由實驗結果可以看出,僅需位置回授的新型控制器在只有位置回授的情形下,效能更勝傳統的計算扭力控制器,而錯誤偵測與判別系統也能夠在系統發生錯誤後的極短時間內,告知使用者系統已經發生錯誤,並判斷出錯誤發生的位置,不僅增加了安全性,也降低硬體成本,增加未來智慧型機器人能夠同時擁有安全與效能兼備的可能性。 This study experimentally evaluates the performance of various kinds of controllers and actuator fault diagnosis system on a two-joint robotic manipulator. On the part of controllers, we use UKF(Unscented Kalman Filter) to estimate (angular) velocities, and compare the performance of the computed-torque controllers with velocities estimated by UKF and backward difference filters respectively. Moreover, we propose a new control method which only requires position feedback. On the part of actuator fault diagnosis system, we combine UKF , GPB-2(General Pseudo Bayesian of Order 2) and IMM(Interacting Multiple Model) algorithms to detect and identify actuator failures of the robotic manipulator. The experimental results show that the new control method has better performance than other controllers. The results also show that the fault diagnosis system can correctly detect and isolate actuator failures in a short time. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT079612520 http://hdl.handle.net/11536/41838 |
顯示於類別: | 畢業論文 |