0
$\begingroup$

I ended up with the following system of equations in the frequency domain: \begin{equation} \mathcal{L}\{f\}(s)=\frac{\alpha (s \beta+ \gamma)}{(s \delta + \alpha+\gamma)(s\beta+\gamma)-\gamma \gamma} \end{equation}

\begin{equation} \mathcal{L}\{g\}(s)=\frac{\alpha \gamma}{(s \beta +\gamma)(s \delta + \alpha+\gamma)-\gamma \gamma} \end{equation}

where $\alpha, \beta, \gamma, \delta$ are constants.

I want to factorise them first so I can transform them back to the time domain.

Can anyone give a hint how to factorise the above equations?

$\endgroup$
1
$\begingroup$

Those are transfer functions of a second order system and integrate as a product of $sin$, $cos$ and $exp$ or $sinh$, $cosh$ and $exp$. You can start from this : $$G(s)=\frac{\alpha\gamma}{s^2\delta\beta+s(\alpha\beta+\gamma\beta+\gamma\delta)+\gamma\alpha}$$ $$=\frac{\alpha\gamma}{(s\sqrt{\delta\beta}+\frac{1}{2}(\alpha\beta+\gamma\beta+\gamma\delta))^2-(\alpha\beta+\gamma\beta+\gamma\delta)^2+\gamma\alpha}$$ Still you need to have some information about the sign of $-(\alpha\beta+\gamma\beta+\gamma\delta)^2+\gamma\alpha$ to use a table of transforms.

If you have no clue about this term the idea to go further would be : $$G(s)=\frac{\alpha\gamma/\delta\beta}{(s+\frac{1}{2}(\alpha\beta+\gamma\beta+\gamma\delta)/\sqrt{\delta\beta})^2-(\alpha\beta+\gamma\beta+\gamma\delta)^2/\delta\beta+\gamma\alpha/\delta\beta}$$ Then let $K=\alpha\gamma/\delta\beta$, $A=\frac{1}{2}(\alpha\beta+\gamma\beta+\gamma\delta)/\sqrt{\delta\beta}$, $\Omega=\sqrt{-(\alpha\beta+\gamma\beta+\gamma\delta)^2/\delta\beta+\gamma\alpha/\delta\beta}$

Note that you have the relation : $$\Omega=\sqrt{K-4A^2}$$ then : $$G(s)=\frac{K}{(s+A)^2+\Omega^2}=\frac{K}{\Omega}\frac{\Omega}{(s+A)^2+\Omega^2}=\mathcal{L}\left[\frac{K}{\Omega}e^{-At}\sin(\Omega t)\cdot u(t)\right]$$ You just have to hope that $\Omega\in\mathbb{R}$ which corresponds to the nice condition $K-4A^2\geq 0$

$\endgroup$
2
  • $\begingroup$ I don't think I have information about that. We can't simplify this any further? $\endgroup$
    – user105627
    Oct 26 '16 at 17:45
  • $\begingroup$ You can certainly simplify but you will have possibly complex expressions in the time domain (which can be transformed into $sin$, $cos$) $\endgroup$ Oct 27 '16 at 8:22
0
$\begingroup$

Factor out the $\beta$ and $\delta$ from the denominator, then you will be left with something of the form $(s+a)(s+b) -c$. Try writing this as $(s-d)^2 - e$. (so I'm working towards 25 in this table.)

Can you work it out from here? You need to write out $(s+a)(s+b)$ and $(s-d)^2$ out and try to find $d$ and $e$ to get the trick done.

$\endgroup$
9
  • $\begingroup$ I can't see the table. The link doesn't seem to be working. Can you please explain it a little bit more. I still can't figure out how to do it. $\endgroup$
    – user105627
    Oct 26 '16 at 10:28
  • $\begingroup$ You are trying to get to the form of transform of $\exp(at) \cosh(bt)$ and $\exp(at)\sinh(bt)$. Generally, when dealing with Laplace transforms it's a bad idea to go guns blazing and factorize your fraction to smaller parts. Try remembering the shape of sin & cos transforms as, those will come about often. $\endgroup$ Oct 26 '16 at 13:13
  • $\begingroup$ @user105627 I added other link to a Laplace table. $\endgroup$ Oct 26 '16 at 13:16
  • $\begingroup$ I thought that by factorising the above it will be possible to end up with something of the form $e^{at}-e^{bt}/a-b$ for the $g$. $\endgroup$
    – user105627
    Oct 26 '16 at 17:42
  • $\begingroup$ You will... in the end. Factorizing is a long road usually. It's way more handy to rewrite in the form of cosh times exp. (note cosh is a sum of two exps itself!). $\endgroup$ Oct 26 '16 at 19:04

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.