Table of Contents
Advertisement
In general, fuzzy logic provides for set definitions that have fuzzy boundaries rather
than the crisp boundaries of Aristotelian logic. These sets can overlap so that, for a
specific input value, one or more sets associated with linguistic labels may be true to
a degree at the same time. As the input varies from the range of one set into the range
of an adjacent set, the first set becomes progressively less true while the second set
becomes progressively more true.
Fuzzy logic has membership functions which emulate human concepts like "tempera-
ture is warm"; that is, conditions are perceived to have gradual boundaries. This con-
cept seems to be a key element of the human ability to solve certain types of complex
problems that have eluded traditional control methods.
Fuzzy sets provide a means of using linguistic expressions like "temperature is warm"
in rules which can then be evaluated with a high degree of numerical precision and
repeatability. This directly contradicts the common misperception that fuzzy logic pro-
duces approximate results — a specific set of input conditions always produces the
same result, just as a conventional control system does.
A microcontroller-based fuzzy logic control system has two parts. The first part is a
fuzzy inference kernel which is executed periodically to determine system outputs
based on current system inputs. The second part of the system is a knowledge base
which contains membership functions and rules.
Figure 9-1
is a block diagram of this
kind of fuzzy logic system.
The knowledge base can be developed by an application expert without any microcon-
troller programming experience. Membership functions are simply expressions of the
expert's understanding of the linguistic terms that describe the system to be controlled.
Rules are ordinary language statements that describe the actions a human expert
would take to solve the application problem.
Rules and membership functions can be reduced to relatively simple data structures
(the knowledge base) stored in nonvolatile memory. A fuzzy inference kernel can be
written by a programmer who does not know how the application system works. The
only thing the programmer needs to do with knowledge base information is store it in
the memory locations used by the kernel.
One execution pass through the fuzzy inference kernel generates system output sig-
nals in response to current input conditions. The kernel is executed as often as needed
to maintain control. If the kernel is executed more often than needed, processor band-
width and power are wasted; delaying too long between passes can cause the system
to get too far out of control. Choosing a periodic rate for a fuzzy control system is the
same as it would be for a conventional control system.
MOTOROLA
FUZZY LOGIC SUPPORT
CPU12
9-2
REFERENCE MANUAL
Advertisement
Table of Contents
