臺灣博碩士論文加值系統

近年來，由於模組控制的高效率性，在工業程序上已被廣泛的應用。
它帶來的好處，對所有的程序控制工程師來說是眾所皆知的。然而一組''
好的''''模式卻是模組控制中不可或缺的重要的因素。
在過去對於多變數系統模式辨識的研究中，大都是以 MISO 或 MIMO
利用 PRBS 的方式對程序產生刺激，最後再將程序輸出的結果配合最小平
方誤差的數值方法，去作程序參數的估算。但利用最小平方誤差法的最大
的問題是，當我們在求取某個程序的參數時，其靜態增益、時延的估算往
往會受到其它程序影響，如此很可能因系統雜訊的干擾而得到錯誤的結果
。尤其當程序為二階或二階以上時，這個問題特別嚴重。
在此提出一個方法，就是將傳統 SISO 回饋繼電器的模式識別方法做
局部的改良。利用 SIMO 的程序輸入方式配合 Lead/Lag 程序來作多變數
系統的參數估算。此一作法的優點是：
1.所需之回饋繼電器測試可以兼顧對程序干擾的程度，並能有效利用測試
所產生的資料。
2.無需進行一次以上的回饋繼電器測試。
3.能確實掌握程序動態的特性。
4.不需引用最小平方誤差方式作參數估算。
5.對於擾動所造成的影響可以有效地適應。

Due to the demand for higher efficiencies, the use of
model-based control in the process industries has proliferated
recently, . Its advantages are well known to all process control
practitioners. However, a ''good'''' model is an essential
ingredient of the model-based control.
Most multivariable system identification methods in the
literature are developed by MISO or MIMO methods using PRBS for
process inputs and least-square is used to find model parameters.
However, the main problem for least-square method is that, when
the model parameters are estimated, the process steady state
gain, delay, etc., are always influenced by other process. If the
system has noise, the result will be lead to the wrong way. When
the process is SOPDT or HOPDT, the problem is especially serious.
In this thesis, an improved SISO relay feedback model
identification method is proposed. This method use SIMO with
Lead/Lag process to estimated the parameters of multivariable
system model. The advantages of the method are :
1.The data obtained from ATV test can reduce the process
disturbance effectively and can make use of the testing data
efficiently.
2.Only one ATV test is enough.
3.The process dynamic can be understood clearly.
4.Least-square is not necessary to estimate process parameters.
5.For the influence of system disturbance, it adapts well.