User Manual

2 - 10
IPN 074-289L
Composer Operating Manual
Or,
 If the change-in-error is negative (moving away from set point), then the
action is large negative (reduce control power by largest allowable limit).
There are about twenty-five such rules in our controller logic. Each rule is
mathematically transformed using operations of fuzzy logic. At the end, the
outcome of all the rule base are combined by centroid method to yield a
numerical value for action. Though the fuzzy control rules can be executed in
real time, it is advantageous to precompute the control action for all possible
values of errors and change-in-errors. Such actions describe the control
surface and can be stored as the fuzzy state-action table. Thus, during actual
control, the computational chore is reduced to mere table look up in real time.
In our experience, a fuzzy controller is easier to set up, approaches set point
faster and settles with minimum overshoot. See Figure 2-4. These are generally
conflicting requirements for a PID controller. The critics of fuzzy logic point out
that the mathematical stability analysis can not be performed due to lack of a
controller model. The advocates of fuzzy logic assert that stability analysis is
neither necessary nor sufficient condition for robustness of an industrial plant
controller. Ultimately, you will have to make the decision as to which control
route is more suitable for the application at hand.