整合資料預測控制與具有延遲及訊息遺失之網路伺服馬達控制系統

Abstract

網路化控制在工業上已經成為一種趨勢，然而，伴隨網路化產生的時間延遲和資料的遺失等問題，輕則降低系統效能，重則使整個系統不穩定。本研究以時間延遲完全補償策略(Perfect Delay Compensation, PDC)處理網路時間延遲的問題。PDC將延遲的影響簡化為原閉迴路系統串接一時間延遲，避免系統不穩定。針對資料遺失的處理以即時轉移機率估測器和智慧型訊息估測器(Intelligent Message Estimator, IME)作資料遺失的補償。即時轉移機率估測器利用即時短時間內訊息遺失的機率快速地計算出轉移機率，成功且有效率的監控資料遺失的分佈情形，並以估測出的轉移機率來切換不同的補償器來補償不同分佈的資料遺失。為了將IME應用於不同的網路控制系統中，提出資料預測控制器(Data-based Predictive Controller, DPC)及時地計算出最符合預期響應之控制訊號，並與即時轉移機率估測器搭配，作為IME補償區間切換的指標。最後結合了DPC、IME與PDC，以不受限於單一系統的架構，同時處理了網路資料遺失與時間延遲的對控制系統產生的影響。經模擬及實驗證明本論文所提出的架構，皆能有效且簡單應用在實際的網路控制系統上，達成網路時間延遲及訊息遺失的補償效果，提升網路控制的效能。

In the network control systems (NCS), both data dropout and time delay greatly affect control performance and stability. In this study, the perfect delay compensation (PDC) is applied to eliminate the network-induced delay effect on the NCS with an AC servo motor, Ethernet, Zigbee, CAN bus. Moreover, applying the least square algorithm and real-time estimation of the transition probability for data dropout, the intelligent message estimator (IME) in a switching mechanism based on the estimated transition probability is developed to reliably estimate the missing message with different dropout distributions. Furthermore, the data-base predictive controller (DPC) is implemented on NCS in this Thesis to predict the best control signals by optimizing system performance. By combining the IME and the predicted signal from DPC, the automatic switching mechanism of IME under different data dropout rates is thus obtained. Finally, the integrated compensator combining all the PDC, IME and DPC leads to satisfactory compensation results to deal with both the time delay and the data dropout in NCS.