0
$\begingroup$

The following proble is from the book:

A Course in Robust Control Theory (a convex approach), middle of p. 200

Consider the following general feedback loop:

enter image description here enter image description here, ie $\dot x(t) = Ax(t) + B_1w(t) + B_2u(t)$
$z(t) = C_1x(t) + D_{12}u(t)$
$y(t) = C_2x(t) + D_{21}2(t)$

Suppose:
1. $v(t) = u(t) - Fx(t)$

Question:
How to show the transfer function: $w$ to $v$ is $S(\hat H,K)$
where,
1. $S(\hat H,K)$ = $H_{11} + H_{12}K(I-H_{22}K)^{-1}H_{21}$
2. $\hat H$ has a state space representation:

$$H = \left[ \begin{array}{c|c c} A & B_1 & B_2 \\ \hline -F & 0 & I \\ C_2 & D_{21} & 0 \end{array} \right]$$

$\endgroup$
1
$\begingroup$

This is an interesting question. Here is how I showed it.

$$\begin{align} \dot{x} &= Ax + B_1 w + B_2 K (C_2 x + D_{21} w) \\ &= (A + B_2 K C_2) x + (B_1 + B_2 K D_{21}) w \\ v &= K(C_2 x + D_{21} w) - Fx \\ &= (K C_2 - F) x + K D_{21} w \end{align}$$

Also, writing down $\hat{H}$, helps

$$\begin{align} H_{11} &= -F(sI - A)^{-1}B_1 \\ H_{12} &= I - F(sI - A)^{-1}B_2 \\ H_{21} &= D_{21} + C_2(sI - A)^{-1}B_1 \\ H_{22} &= C_2(sI - A)^{-1}B_2 \end{align}$$

Now, we need to show that $S(\hat{H},K)$ equals to

$$T_{wv} = K D_{21} + (K C_2 - F)(sI - A - B_2 K C_2)^{-1} (B_1 + B_2 K D_{21}) $$

The main idea in here is to use the Matrix Inversion Lemma to obtain

$$ ([sI - A] + [- B_2 K] I [C_2])^{-1} = (sI - A)^{-1} + (sI - A)^{-1} B_2 K (I - C_2 (sI - A)^{-1} B_2 K)^{-1} C_2 (sI - A)^{-1} $$

Now, we just need to manipulate $T_{wv}$ and replace $H_{ij}$ in proper places, which is easy but time consuming so I will skip this part. I couldn't find a clever way to overcome the calculations.

$\endgroup$
  • $\begingroup$ All papers and books I have seen always say "it is routine to show...(w/o showing)". In fact these derivations are not quite trivial. $\endgroup$ – sleeve chen May 25 '15 at 22:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.