I want to design a robust control system using the internal model design specifications. The block diagram is the one shown below:

enter image description here

I am trying to obtain the transfer function $\ \frac{Y(s)}{R(s)} $ in order to acquire the characteristic polynomial of the closed loop system but for some reason I am stuck. This is what I have done so far:

$$ E_{a}(s) = R(s) - Y(s) $$

$$ U(s) = E_{a}(s)G_{c}(s) - KX(s) $$ ( $\ U(s) $: control input to the process $\ G(s) $)

$$ Y(s) = U(s)G(s) = E_{a}(s)G(s)G_{c}(s)-KX(s)G(s) $$

$\ $

$$ Y(s) = R(s)G(s)G_{c}(s) - Y(s)G(s)G_{c}(s) - KX(s)G(s) $$ $\ $

$$ Y(s) = \frac{G(s)[R(s)G_{c}(s)-KX(s)]}{1+G(s)G_{c}(s)} $$

From this point, I really don't know how to continue. Obviously the term that confuses me is $\ X(s) $. So, now how should I proceed and obtain the overall transfer function ?

  • $\begingroup$ Diadikasia. I love Greek :-) $\endgroup$ – peterh - Reinstate Monica Mar 27 '20 at 11:09
  • $\begingroup$ Yes, I like Greek too 😀 But what about the transfer function ? Any ideas ? :-) $\endgroup$ – Teo Protoulis Mar 27 '20 at 12:41
  • $\begingroup$ There was a deleted answer: "I think the problem is that you are not considering that X(s) = Y(s)." It was deleted on obvious reasons, but maybe it helps a little bit. $\endgroup$ – peterh - Reinstate Monica Mar 27 '20 at 12:48
  • $\begingroup$ Well $X(s)$ consists of the states of the system (position,velocity) and $Y(s)$ is just the output of the system which is position. I don’t think this is a correct approach. $\endgroup$ – Teo Protoulis Mar 27 '20 at 12:49
  • 2
    $\begingroup$ But that would be the transfer function of an output feedback, we have a state feedback. I think you need to write the transfer function in some state space fashion, involving A, B, C and D, to describe how x and y differ. y = Cx, so your approach only works if C is filled with 1s. Is there no more info on the internals of G(s)? $\endgroup$ – OpticalResonator Mar 27 '20 at 13:43

Well, after a while this is what I finally came up with. I have tested the results in simulation environment as well as on the real system and the produced control behaviour was what I expected. Firs of all, I rewrote the block diagram in the form shown below:

                                        enter image description here

The whole procedure of the transfer function is shown below:

The state variables of the plant process are $\ x_1, \ x_2 $ and they are equal to:

$$ x_1 = y \rightarrow X_1(s) = Y(s) $$ $$ x_2 = \dot{y} \rightarrow X_2(s) = sY(s) - y(0) $$

Considering zero initial conditions, the transfer function of the inner loop is (simple feedback system):

$$ H(s) = \frac{G_p(s)}{1+k_1G_p(s)+sk_2G_p(s)} $$

And now it all comes down to the outer loop being again a simple feedback loop:

$$ T(s) = \frac{G_c(s)H(s)}{1+G_c(s)H(s)} $$


In recent comments, you have attached some important info that you missed to put in your question, primarily that there are two states: position $y=x_1$, and velocity $x_2$ and the output is $x_1$. In that sense $X(s)\neq Y(s)$. An image of your block diagram can be represented as

enter image description here

That means that the matrix $C$ is different from the Identity matrix. Remember that $\dot{x_1}=x_2$ and that you can consider your plant as $\dot{x}=Ax+Bu$.

You are probably trying to obtain the state-space equation. I'd say that your problem also has important info about Gc(s).


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.