Specifying rule processing options
There are four optional statements that, if used, must appear
immediately after the RuleSet statement. The
statements may appear in any order, and if any statement is
entered more than once, the last one takes precedence. These
statements give you some control over the processing of the
fuzzy inference engine while it is evaluating your rules.
Specifies the threshold at which truth values become
insignificant. If the truth value of any antecedent clause falls
below the alphacut when the clause is evaluated, the inference
engine stops evaluating the rule in which the clause appears.
If left unspecified, the default AlphaCut value is 0.1.
Syntax
AlphaCut ( <numericValue> )
Parameters
- <numericValue>
- Is a number from 0.0 to 0.99. If set at 0.0, any clause that
evaluates to anything greater than zero is essentially
considered true to some degree. If set at 0.99, all clauses must
evaluate to 1.0 (boolean true) to be considered true at
all. Values between 0.05 and 0.2 may be considered typical, but
of course it all depends on your fuzzy variables and their fuzzy
set definitions.
Example
AlphaCut ( 0.20 )
Return to top
Specifies the manner in which a rule's consequent fuzzy region
is correlated with the rule's antecedent fuzzy truth value.
If left unspecified, the default CorrelationMethod is Product.
Syntax
CorrelationMethod ( Product | Minimum )
Parameters
- Product
- Specifies that the membership value of the consequent fuzzy
region is the product of the fuzzy region and the truth of the
premise. The effect is that the consequent fuzzy region is
scaled, preserving its shape.
This method is generally used with InferenceMethod(ProductOr) or
InferenceMethod(FuzzyAdd).
- Minimum
- Specifies that the membership value of the consequent fuzzy
region is the minimum of the fuzzy region and the truth of the
premise. The effect is that the consequent fuzzy region is
truncated at the truth of the premise, creating a plateau.
This method is generally used with InferenceMethod(MinMax).
Example
CorrelationMethod ( Minimum )
Return to top
Specifies the manner in which a fuzzy set is turned into a crisp
numeric value.
If left unspecified, the default DefuzzifyMethod is Centroid.
Syntax
DefuzzifyMethod ( Centroid | MaxHeight )
Parameters
- Centroid
- Specifies that a fuzzy set's crisp value is calculated as
the weighted mean (center of gravity) of the fuzzy region. Also
known as composite moments.
- MaxHeight
- Specifies that a fuzzy set's crisp value is calculated from
the point that has the highest truth value. Also known as
composite maximum.
Example
DefuzzifyMethod ( Centroid )
Return to top
Specifies the manner in which a solution variable's fuzzy
representation is updated.
If left unspecified, the default InferenceMethod is FuzzyAdd.
Syntax
InferenceMethod ( FuzzyAdd | MinMax | ProductOr )
Parameters
- FuzzyAdd
- Specifies that the fuzzy solution set is updated by adding
the minimum truth value of the consequent fuzzy region, bounded
by 1.0.
This method is generally used with CorrelationMethod(Product).
- MinMax
- Specifies that the fuzzy solution set is updated by using
the maximum of the minumum truth value of the consequent fuzzy
set.
This method is generally used with CorrelationMethod(Minimise).
- ProductOr
- Specifies that the fuzzy solution set is updated by using 1
minus the product of (1 minus the truth value of the consequent
fuzzy set) and (1 minus the predicate truth).
This method is generally used with CorrelationMethod(Product).
Example
InferenceMethod ( FuzzyAdd )
Return to top
Rule language table of contents.
Fuzzy editor table of contents.
Fuzzy System table of contents.
Last modified: Fri Feb 18 10:58:01 CST 2000