標題: | 使用快速GPB-2演算法的即時機器人致動器錯誤偵測與識別 Real-Time Detection and Identification of Robot Actuator Failures with Fast GPB-2 Algorithm |
作者: | 黃暉鈞 Huei-jyun Haung 蕭得聖 Te-Sheng Hsiao 電控工程研究所 |
關鍵字: | 錯誤偵測;錯誤識別;致動器;fault detection;fault isolation;actuator |
公開日期: | 2007 |
摘要: | 本論文的貢獻在於對機械臂僅裝置角度編碼器以降低成本的情況下,發展應用於機械臂的即時錯誤偵測與辨識演算法,以期能對機械臂的運作狀態進行良好估測,增加機器人運作的安全性。文中將使用非追蹤型卡曼濾波器進行機械臂角度和角速度的估測,使用期望值最大化演算法估測摩擦力,利用濾波扭力計算與多模型組態GPB-2演算法來判斷錯誤模型的種類,由模擬結果可看到該即時演算法的確比原始的多模型組態GPB-2演算法更為精確與有效率。 This thesis has a great contribution on detection and identification of the robot's actuator failures by combining UKF(Unscented Kalman Filter), GPB-2 algorithm, EM algorithm and pre-classification of fault models. Simulation results show that the proposed FDI(fault detection and identification) algorithm is better than conventional GPB-2 algorithm in terms of computational efficiency and accuracy of fault detection. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT009512528 http://hdl.handle.net/11536/38241 |
顯示於類別: | 畢業論文 |