Fuzzy PD controller

Carlo Pangrazio (carlopan@numerica.it)
Mon, 17 Nov 1997 03:08:54 +0100 (MET)

Hi!
I'm working on a fuzzy PD controller based on uC80535-INTEL (8 bit uC).
The program uses a 24 bit floating-point arithmetic.
The simulations showed a different performance of the control variable,
depending on the input combination method used: if we use the product
if-part method, the control variable has the same shape we can obtain
during a PC simulation (based on a 64 bit floating-point arithmetic) on
the same data input; instead, if we use the min if-part method, even if
the system reaches the set-point at the same time, the control variable
is more oscillating. So it seems that the product if-part method is
robust as regards the floating-point format imprecision.
Can anyone suggest to me some papers on the question?
Another problem: which is the exact definition of FLIPs?
FLIPs = n of rules processed/sec
or
FLIPs = n of fired rules processed/sec ?

Thanks advance

Carlo Pangrazio
Brescia
Italy
carlopan@numerica.it