Page 32 - Practical Control Engineering a Guide for Engineers, Managers, and Practitioners

Qualitative Concepts in Control Engineering 7

0.5 .-----.--....---....----.-----.-----.-----:-----.---r----.----,

0.4
0.3

0.2
Qj 0.1
~
~ 0
~ -0.1

-0.2
-0.3

-0.4
-0.5~~5 7 0--~~~~~~~~~~~~~~~~500
Time
F1oURE 1-10 Autocorrelated stochastic variation of the hotel tank level.

When the flushing of the toilets and the usage of the faucets in the
rooms is completely unpredictable and independent of each other, the
tank level variation might look like Fig. 1-9. For the time being we will
refer to these kinds of fluctuations as wumtocorrelated stochastic distur-
bances where the word "stochastic" means conjectural, uncertain, or
unpredictable. We will avoid using the word "random" because of the
many confusing connotations. Also, we will defer the definition of
"unautocorrelated" until a later chapter. If the stochastic variation is
autocorre/ated, the hotel tank level might look like Fig. 1-10.
Later on in Chap. 8, a significantly more quantitative definition
will be attached to these two kinds of disturbances and we will find
out how to characterize them. For the time being, suffice it to say that
unautocorrelated disturbances are stochastic variations with a con-
stant average value while autocorrelated disturbances exhibit drift,
sometimes with a constant overall average and sometimes not.

1-4 Comparing Feedforward and Feedback Controllers
The feedforward controller can act on a measured event (such as the
drain value position) before it shows up as a disturbance in the process
output (such as the tank level). Unfortunately, the feedforward controller
has no idea how well it did. Furthermore, it is often rather difficult to
measure the disturbance-causing event. Sometimes there will be many
disturbance-causing events, some of which cannot be measured. Also, it