適應性干擾觀測器與適應性預測干擾觀測器應用於一階穩定時延系統消除未知的確定性干擾

Abstract

本論文提出適應性干擾觀測器(Adaptive Disturbance Observer, ADOB)與適應性預測干擾觀測器(Adaptive Predictive Disturbance Observer, APDOB)，消除非時延系統與時延系統中的未知確定性干擾，此架構是基於適應性控制中的自適應性控制器的概念(Self- Tuning Regulator, STR)，結合干擾觀測器或預測干擾觀測器與最小平方遞迴遺忘因子法(Recursive Least Squares with exponential forgetting)，估測系統的參數與控制器Q的參數r，藉由更新參數，克服模型不確定性與消除未知的確定性干擾，當系統有時間延遲不確定性時，增加可變迴歸估測法(Variable Regress Estimation, VRE)，消除時間延遲不確定性。為了調整系統的伺服響應，所以在非時延系統的外迴路增加PI控制器，在時延系統的外迴路增加修改型史密斯預測器(Modified Smith predictor)。本文所提出的架構對於一階穩定非時延系統與一階穩定時延系統能消除未知的確定性干擾與模型不確定性且維持良好的輸出響應，最後將控制架構應用於永磁無刷線性導軌，考量受到量測誤差的影響，因此修改控制架構，以補償系統所受到的干擾及不確定性。

This thesis proposed Adaptive Disturbance Observer (ADOB) and Adaptive Predictive Disturbance Observer (APDOB) to reject unknown deterministic disturbance of non-delay system and delay system. This structure is designed based on one of the concepts of adaptive control, Self-tuning Regulator (STR), and constructed by combining Disturbance Observer or Predictive Disturbance Observer with Recursive Least-Squares with exponential forgetting to estimate the system parameters and parameters r of controller Q. The system can overcome model uncertainty and reject unknown deterministic disturbance by updating parameters. Apply Variable Regress Estimation (VRE) to eliminate time delay uncertainty when the system has time delay uncertainty. In order to adjust the servo performance of the system, we add PI controller in outer loop for non-delay system and Modified smith predictor in outer loop for delay system. A control structures proposed by this thesis can reject unknown deterministic disturbance and model uncertainty to maintain a good output performance for stable first-order non-delay system and stable first-order delay system. Finally, the control structure can apply to the P.M. linear motor. The problem of measurement noise must be concerned. Therefore, we modified the control structure to compensate the uncertainty and disturbance.