統一預測控制應用於一階穩定時延系統

Abstract

中文摘要 在運動控制領域中因為實務上的需求，時延系統、模型不確定性及干擾的消除皆被廣泛的研究。為了處理上述的問題，本文分兩部分各別做討論；第一部分(基於強健穩定的概念)：將統一預測控制(Unified Predictive Control, UPC)應用於受到確定性干擾且包含模型誤差的一階穩定時延系統，並分析該系統之強健性及其標稱性能(nominal performance)，最後提出一個實用且簡單的參數調整方法。此部份的結果能確保系統為強健穩定，並在強健穩定的條件下擁有最佳的標稱性能。第二部分(基於適應性控制的概念)：提出一個適應性統一預測控制架構(Adaptive Unified Predictive Control, Adaptive-UPC)應用於一階穩定時延系統以消除模型不確定性及未知的確定性干擾。這個新架構是基於適應性控制中間接自調變器(Indirect Self-tuning Regulator)的概念，結合統一預測控制、含有遺忘因子的遞迴最小平方估測法(Recursive Least-Squares Estimation with forgetting factor, RLS with forgetting factor)與可變回歸估測法(Variable Regression Estimation, VRE)而成的。此部分的結果能有效的應用在一階穩定時延系統，並能準確估測出程序參數（增益、時間常數及時間延遲）與消除未知的確定性干擾。在實務的情況下必須考慮量測雜訊帶來的影響，因此在實作部份提出一改良型的適應性統一預測(Modified Adaptive Unified Predictive Control, Modified Adaptive-UPC)控制來降低雜訊的影響。

In the field of the motion control, dead-time process with model uncertainty and disturbance are widely studied because of the practical importance. This thesis is divided into two parts to deal with these problems, respectively; Part I (based on the concept of robust control): One applies Unified Predictive Control (UPC) to first-order-plus-dead-time (FOPDT) stable processes with concerns of model uncertainty and deterministic disturbance, and analyzed the robustness and nominal performance of the system. A simple and useful parameter tuning method is proposed. The results of this part guarantees optimal nominal performance based on robust stable condition. Part II (based on the concept of adaptive control): one proposes an Adaptive Unified Predictive Control (Adaptive-UPC) and applied it to first-order-plus-dead-time (FOPDT) process model to eliminate effects of model uncertainties and unknown deterministic disturbance. This new control structure is designed based on one of the concepts of adaptive control, Indirect Self-tuning Regulator, and constructed by combining UPC controller, Recursive Least-Squares Estimation with forgetting factor (RLS with forgetting factor) and Variable Regression Estimation (VRE). The results of this part can be effectively used to first-order-plus-dead-time (FOPDT) process models and can accurately estimate the process parameters (gain, time constant and dead-time) with elimination of unknown deterministic disturbance. For practical purpose, the problem of measurement noise must be concerned. Therefore the Modified Adaptive Unified Predictive Control (Modified Adaptive-UPC) is proposed to decrease the effect caused by measurement noise.