[Scilab-users] Question syslin

[Andreas Ladanyi]

> On Tue, 2014-12-16 at 11:01 +0100, Andreas Ladanyi wrote:
>> Hi,
>>
>> i am reading the help of scilab to understand the syslin function. In
>> the help i can see 3 options for the dom parameter. There is an option
>> for a sampled system 'n'.
>>
>> In this help there is no example for a sampled system. So what is the
>> difference between a discrete system and a sampled system ? What is the
>> difference between an discrete system d and an sampled system n for syslin ?
>>
>> In the most examples i read in the web syslin is called with dom='c'.
> Hey Andreas:
Hi Tim,

thank you for your excellent description. Could this be inserted to the 
online help of the syslin function ?

cheers,
Andreas
> Domain = 'c' is for a continuous-time system, where the system is
> defined as
>
> dx/dt = A * x + B * u,
> y = C * x + D * u
>
> This is a normal continuous-time linear state-space system description.
>
> Domain = 'd' is for a discrete-time system where the sampling interval
> is left undefined for whatever reason (in my case, it's usually because
> I'm being lazy, but sometimes it's because the sampling interval isn't
> constant, or because there's no meaningful "sampling interval" in the
> problem).
>
> It defines the sampled-time system:
>
> x{k} = A * x{k-1} + B * u{k}
> y{k} = C * x{k-1} + D * u{k}
>
> (Note the mixed time indexes on the input and state variables on the
> right-hand side of these equations.  You'll sometimes see this expressed
> differently, so if you're trying to implement something from an article
> or book, pay attention!)
> Domain = n is for a discrete-time system where the sampling interval is
> defined.  It works exactly like domain = 'd', except that things that
> depend on the real-world frequency, like Bode plots, will come out
> right.
>
>      
> When I am doing control system design this is usually the form that I
> use, because by the time I'm down to this level of detail I've usually
> established the sampling rate, and I'm working at tuning the system to,
> or verifying it against, some real-world criteria that must be expressed
> in the frequency domain.
>