• 沒有找到結果。

以LMI 方式設計離散時間無限維系統之控制器

N/A
N/A
Protected

Academic year: 2021

Share "以LMI 方式設計離散時間無限維系統之控制器"

Copied!
4
0
0

加載中.... (立即查看全文)

全文

(1)

1

行政院國家科學委員會專題研究計畫成果報告

以 LMI 方式設計離散時間無限維系統之控制器

Contr oller Design for Discr ete-Time Infinite Dimensional Systems

Using the LMI Appr oach

計畫編號:NSC 88-2213-E-002-089

執行期限:87 年 8 月 1 日至 88 年 7 月 31 日

主持人:馮蟻剛 臺灣大學電機系

一、中文摘要 在本計畫中討論無限維參數系統在邊 界控制下的強健穩定性問題。對於控制訊 號由邊界進入之無限維系統,提出一個處 理其強健穩定性的分析方式。 此一分析方式首先將邊界控制問題轉 換成一個在內插空間上的歌西問題,使得 一般狀態空間的分析方式得以引入。 對於幾類具有特殊形式的無限維參數 系統,本計畫亦提出了較容易的計算方 式,並給予一個例子以說明所推導理論的 計算方式。 關鍵詞:無限維參數系統,邊界控制,強 健穩定性 Abstr act

In this report, we consider the stability robustness problem for infinite dimensional systems with boundary control inputs.

We first translate the problem into a standard Cauchy problem in certain interpolation space. In such an interpolation space, the boundary control robustness problem can be represented as a standard Cauchy problem with zero boundary conditions, therefore some standard stability robustness analysis methods can be modified to find the robustness bounds..

For certain types of infinite dimensional systems, a practical computation method is

developed so that numerical results are easier to obtain.

Keywor ds: infinite dimensional systems,

boundary control problem, stability robustness 二、緣由與目的 關於無限維參數系統的研究,有相當多 的研究方向是受有限維系統中的結果所啟 發。在半群理論(Semi-group Theory)的發 展下,的確看到無限維參數系統有許多性 質是與有限維系統的表現相當接近[5,8]。 但探究由偏微分方程所描述的無限維系 統,則可看出邊界值問題是常微分與偏微 分方程所描述系統間的最大差異。 在由偏微分方程所描述的系統中,外界 的能量交換通常是在邊界上進行,也因 此,邊界控制問題一直是在實際物理及工 程應用上所必須面對的問題。 在[7]的研究中,曾經討論了特定無限 維參數系統的強健穩定性問題,但所討論 的系統侷限在沒有邊界控制的情況下,也 因此並無法解決控制信號需由邊界進入的 情況。 關於無限維參數系統的邊界值問題,文 獻上可看到由 Lion[6]等人在 60 年代中所 提 出 關 於 跡 (Trace) 及 跡 算 子 (Trace Operator)的處理方式,詳細討論並引入了 內插空間(Interpolation Space)以解決此種 邊界值的問題。 在本篇報告中,我們參考由 Amann[2] 等關於邊界問題的處理模式,巧妙的將邊

(2)

2 界值問題轉換為在內插空間上不具邊界輸 入的問題模式。這種問題形式則可直接應 用 於 在 [7] 所 導 出 關 於 強 健 穩 定 性 的 結 果,而直接處理具有邊界控制項的強健穩 定性問題。 三、研究方法與成果    令 Z 為一希伯特空間(Hilbert Space), Z Z A D A0 : ( 0)⊂ → 為一定義在 Z 上的無 界算子(Unbounded Operator),D(A0)在 Z 中 為 稠 密 (Dense) , 且 為 閉 算 子 (Close Operator) 。 由 A 產 生 出 的 線 性 半 群0 (C Semigroup)為0 T0(t)。考慮如下之系統        = = = + = 0 0 ) 0 ( ) ( ) ( ) ( ) ( ) ( ) ( z z t Cz t y t g Ez t Bu t z A t z dt d (1) 其 中 BL(U,Z),CL(Z,Y) 為 有 界 算 子 (Bounded Operator) , E 為 邊 界 算 子 (Boundary Operator)。u(t)及g(t)為控制輸 入訊號,y(t)為輸出訊號。並假設(1)所描 述 的 系 統 為 指 數 穩 定 (Exponentially Stable)。定義A 的相對有界算子(Relative0 Bounded Operators)J(A0)為 J(A0)={A|D(A0)⊂ D(A), Azα A0z +β z } 考慮擾動算子 AiJ(A0),i =1,...,N0,及以 下的受擾動系統:         = = = + + =

− 0 1 0 ) 0 ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( 0 z z t Cz t y t g Ez t Bu t z A k A t z dt d N i i i (2) 令u(t)=0,g(t)=kN0+1y(t),即只考慮由邊 界所提供的輸出回受控制,整理得如下之 包含擾動項之輸出迴授控制系統:        = ⋅ = + = + −

0 1 1 0 ) 0 ( ) ( ) ( ) ( ) ( ) ( 0 0 z z t Cz k t Ez t z A k A t z dt d N N i i i (3) 注意在ki =0,i =1,...,N0 +1下,由(3)所描述 的系統為指數穩定。 令

[

]

T N k k k= 1,..., 0+1 ,考慮由(3)所描述 的系統,則一邊界強健穩定性問題即是找 到 k 的一個上界 k 使得當 k <k時由(3)所 描述的系統仍是指數穩定。 在[7]中曾經討論了在沒有邊界輸入之 下的強健穩定性問題,而在具有邊界控制 項下,我們引入在內插空間(Interpolation Space)下的表達式,以處理邊界控制問題。 定義 1 若XY, X, 為 Hilbert 空Y 間 , A:XY為 正 定 自 伴 算 子 (Positive and Self-Adjoint Operator) , 對 任 意

1 0≤θ, 定 義 X 與 Y 之 間 的 內 差 空 間 為: ) ( ] , [X Y1θ =D (4) 在內差空間的定義下,我們可由A 建0 構出一組內插空間A0,s。首先,對於整數 0 ≥ s 定義 ( ( 0), ) s s s D A H = • ,而對於s<0 定 義 Hs 為 ( ( 0), ) s s A D • 所 完 備 化 (Completion)之空間。接著,對於s≥0令 s A0, 為A 在0 H 下的實現(Realization) ,而s 0 < s 時 則 為 A 在0 H 的 閉 延 拓 (Closeds Extension) 。 再 令 Hα =[Hs,Hs+1]αs, Z s s s<α < +1, ∈ ,及A0,αA0,sH 下α 的實現s<α <s+1,sZ。在此程序下,我 們 有 了 一 組 由 A 所 建 構 出 的 函 數 空 間0 } ), , {( A0,α α∈ℜ,且每一A 產生一線性0,α 半群 A t e t T α α( ) 0, , 0 = 。 有 了 這 組 由 A 所 產 生 的 函 數 空 間0 } ), , {( A0,α α∈ℜ ,令ρ(A0)為A 之預解0 集(Resolvent Set),我們引入系統(3)在內插 空間中的表達式。

(3)

3 定理 2[2] 給定一σρ(A0),則存在 一β∈(0,1)及一線性算子R ,使得由(3)所β 描述的系統與下述之系統       = − + + = − + − − −

0 1 , 0 1 1 1 , 1 , 0 ) 0 ( ) ( ) ( ) ( ) ( 0 0 z z z C R A I k t z A k A t z dt d N N i i i β β β β β σ (5) 在由Hβ1所決定的測試函數(Test Function) 下其弱解(Weak Solution)是一致的。 以下的引理是敘述主要結果所必須 的: 引理 3 (σIA0,β−1)RβCβ為一A0,β−1之 相 對 有 界 算 子 , 即 (σIA0,β1)RβCβ ∈ ) (A0,β1 J , 且 其 相 對 有 界 係 數 (Relative Bounded Coefficient) 為 β β β α = R C , β β β σ β = ⋅ R C 。 下面的定理是主要的結果: 定 理 4 若 由 (5) 所 描 述 的 系 統 在 0 = i k 時為指數穩定,且為正規算子,亦即 有 µ,ϖ >0 使 得 T t e ϖt β β−1 1µ − , 0 ( ) 。 令 zdt T T z P 0 0, 1 * 1 , 0 1 , 0 − ∞ − − =

β β β ,且令 1 1 1 , 0 * 1 , 1 , 1 , 0 1 , ) ( inf − − − − − − = + = β β β β β A A P z z P k i i z i 1 1 2 1 1 (2(1 ) ) 0 0 0 − + + + = + N + N N r k β ϖ µ α 其中 1 1 , 0 1 , 0 − − = A β P β β r ,令 1/2 1 1 2 ) ( 0 − + = −

= N k i k k 則當 k <k,由(3)所描述的系統仍為指數 穩定。 定理 4 所描述的為一般性的結果,此 結果在實際的計算上不太方便,故而對較 特殊的A 我們給出下面的結果:0 系 理 5 若 A0,β1 為 自 伴 算 子 (Self Adjoint),且對每一個擾動Ai,β1其相對有 界係數α , 均為已知,則i βi 1 ) ( + − ≥ β ϖ µ αi i k , i=1,...,N0 +1 若對所有的擾動的相對有界係數皆是 已知,則類理 5 的結果提供了一個相當簡 單估計強健穩定範圍的方式。下面是一個 說明的例子。 範例 6 考慮如下的熱傳導方程式:            =         =           ∂ ∂ ∂ ∂ − ∂ ∂ = ∂ ∂

0 2 1 1 2 2 ) 0 ( ) , ( ) ( ) , ( ) ( ) 1 , ( ) 0 , ( ) , ( ) 1 ( ) , ( z z dx z t z x g dx z t z x g k t t z t t z x t z x x t t z (6) 很容易可以檢查當k1 =0時此系統為指數 穩定。給出相對應之算子: ) 1 ( 2 2 0 − ∂ ∂ = x A           ∂ ∂ ∂ ∂ = ) 1 , ( ) 0 , ( t t z t t z Ez ,         =

g x z t z dx dx z t z x g Cz ) , ( ) ( ) , ( ) ( 2 1 選 擇σ =0∈ρ(A0), 4 3 2 1 < <β , 並 選 取 2 1,g g 使得 C β =1,則有 β β R k1 ≥ 1 故當 1 1 k k < 由 (6) 所 描 述 的 系 統 仍 為 指 數 穩 定。 四、結論與討論 在本報告中提出了一個處理包含邊界 迴授控制無限維參數系統的強健性問題分 析處理方式。對特定形式的系統及擾動亦 給出一較為簡易的計算方式。報告中並包 含一個範例作為說明。

(4)

4 五、計畫成果自評 本計畫雖以離散時間無限維系統做為 開始之研究對象,但在研究過程中,發現 無限維參數系統之邊界控制問題十分特 別,且具有其理論重要性,因而轉移研究 目標。在新的研究目標中,我們發展出一 套新的理論,使邊界控制之處理方式更為 系統化。此部分結果目前正投稿國際期刊 中。 六、參考文獻

[1]Ahmed, N. U. and X. Xiang, “Nonlinear Boundary Control of Semilinear Parabolic Systems,” SIAM J. Contr. Optim., vol. 34, no. 2, pp. 473-490, 1996.

[2]Amann, H., “Parabolic evolution equations and nonlinear boundary conditions,”J. Differ. Equations, vol. 72,

pp. 201-269, 1988.

[3]Bensoussan, A., G. D. Prato, M. C. Delfour, and S. K. Mitter, Representation and Control of Infinite Dimensional Systems, Volume I, Birkhäuser, Boston,

1992.

[4]Conway, J. B., A Course in Functional Analysis, Springer-Verlag, New York,

1990..

[5]Curtain, R. F. and H. Zwart, An Introduction to Infinite-Dimensional Linear System Theory, Springer-Verlag,

New York, 1995.

[6]Hinrichsen, D. and A. J. Pritchard, “Robust stability of linear evolution operators on Banach spaces,” SIAM J. Contr. Optim., vol 32, no. 2, pp.

1503-1541, 1994.

[6]Lion, J. L., and E. Magenes, Non-Homogeneous Boundary Value Problems and Applications, Volume I,

Springer-Verlag, New York, 1972.

[7]Lu, S. H. and I-K. Fong, “Stability radius of linear normal distributed parameter systems with multiple directional perturbations”, Proc. 37th IEEE Conference on Decision and Control,

pp.819-820, 1998

[8]Pazy, A., Semigroups of Linear Operators and Applications to Partial Differential Equations, Springer-Verlag,

New York, 1983.

參考文獻

相關文件

• Adding restrictions on the allowable solutions (the solution space) may make a problem harder, equally hard, or easier.. • It is

• Adding restrictions on the allowable solutions (the solution space) may make a problem harder, equally hard, or easier.. • It is

• To the right of the Draw mode buttons you find push buttons through which you can access all the functions that you need to define and solve the PDE problem: define

With the proposed model equations, accurate results can be obtained on a mapped grid using a standard method, such as the high-resolution wave- propagation algorithm for a

數位計算機可用作回授控制系統中的補償器或控制

If the subset has constant extrinsic curvature and is a smooth manifold (possibly with boundary), then it has an explicit intrinsic lower curvature bound which is sharp in

Write the following problem on the board: “What is the area of the largest rectangle that can be inscribed in a circle of radius 4?” Have one half of the class try to solve this

By this, the second-order cone complementarity problem (SOCCP) in H can be converted into an unconstrained smooth minimization problem involving this class of merit functions,