Introduction to GAF
GAF (Genetic-Adapt Fuzzy Control) runs in IBM PC and compatible machines.
The development of GAF is intended for off-line simulation, adaptation
and on-line adaptive control. Currently the functions for simulation and
off-line adaptation are completed, the on-line portion is still under
GAF builds run time data directly from fuzzy IF-THEN rule segments
contained in English like text files. It eliminates intermediate
compilation and achieves quick turn-around time.
A segment, a basic unit in GAF, can be defined by means of:
combination of fuzzy rules and math formulas
Not only the control can be defined in segments, the feedback and the
evaluation can also be defined for simulation and adaptation. In future
release GAF will be able to use data set (measured or modeled) as its
With GAF's graphic display and user interface it also provides user as an
education tool for understanding fuzzy logic and genetic algorithm.
GAF allows users to generate a fuzzy control system by simply defining
the inputs, outputs, data set, and initial rule sets. GAF uses genetic
algorithm to derive proper rules and fuzzy sets from the initial rules.
By changing, adding, deleting rules and fuzzy membership sets, the genetic
algorithm automatically adapts and optimizes the fuzzy control system.
GAF provides an integrated simulation environment for user to fine-tune
their fuzzy control applications and examine the response of the fuzzy
rules with certain conditions (i.e. for some known input values). High
lights of GAF's simulation environment are:
Verify single segment
Change schedule rate
Single step to view details of fuzzy inference
Change output gain
GAF's Off-line Adaptation
With evaluation segment, GAF is capable of adapting user's fuzzy control
applications automatically. In future release GAF will provide user with
canned evaluation method. Major adapting functions are:
Change existing rule
Change fuzzy membership set
Adding new rule
Disable existing rule
Alter cycle time
Alter output gain