Re: Sampling time

Jan Jantzen (jj@iau.dtu.dk)
Sat, 22 Feb 1997 15:19:22 +0100


On Fri, 21 Feb 1997, Ali Khalaf wrote:
>
> In some of the literature , the fuzzy logic controller uses the error
> e(k) and the CE=e(k)-e(k-1) . In others fuzzy logic controller uses the
> error e(k) and CE=(e(k)-e(k-1))/Ts , where Ts is the sampling time .
> What are advantages and disadvatges of both approaches .

The straight difference, the first approach, is how they did
it in the old days. The second approach is better, because
then tuning of the gains becomes independent of the choice
of sampling period.

Similarly with integrators. If for example the controller
output is CHANGE in output, CU, and if you compute U = SUM(CU*Ts)
then you can tune the output gain GCU independently of the
choice of sampling period Ts.

Regards, Jan

Technical University of Denmark | Tel +45 4525 3561 (direct)
Att: Jan Jantzen | Fax +45 4588 1295
Dept. of Automation, Bldg. 326 | Jantzen@iau.dtu.dk
DK-2800 Lyngby, DENMARK | http://www.iau.dtu.dk/~jj