For example, the shaded block corresponds to the rule: "When the error is positive large and its derivative is zero, make the input positive large." From this, several problems with designing fuzzy controllers are evident. First, the notion of stability was never addressed. The designer simply uses experience and intuition to determine what "large" means and what the rule base should be. Second, the complexity of the rule base goes up as mn where m is the number of fuzzification levels and n is the number of state variables. Third, the examples presented in the literate were for second-order
systems which are controllable through a PID compensator.
The development of a rule base for higher-order unstable systems, where PID controllers fail completely is not straightforward.
Note from Table 1 that a symmetry exists. Similar symmetry is also present in other fuzzy rule bases such as the 7x7 rule base used in Hwang [6] or 9x9 used in Bouslama [10]. This suggests that what is really being measured is the distance from the diagonal. Noting this allows the fuzzy rule-base to be simplified as follows:
.
systems which are controllable through a PID compensator.
The development of a rule base for higher-order unstable systems, where PID controllers fail completely is not straightforward.
Note from Table 1 that a symmetry exists. Similar symmetry is also present in other fuzzy rule bases such as the 7x7 rule base used in Hwang [6] or 9x9 used in Bouslama [10]. This suggests that what is really being measured is the distance from the diagonal. Noting this allows the fuzzy rule-base to be simplified as follows:
.