Information Technology Reference
In-Depth Information
The population is generated randomly. Each real number, representing the
value of a parameter, is encrypted in GAs as a binary string. An example of
encodingnumberasabinarystringispresentedinFig7.
Fig. 7.
Exampleofrepresentationofnumbersasbinarystrings
Each individual of a population is characterized by its fitness. The fitness
function represents how good is the solution of the problem for given set of
parametersvaluesofanindividual.Inthedevelopedalgorithmtheauthorspro-
posedthefitnessfunction(3)tofindthebestsolution.
i
exp
))
2
f
b
(
i
)=(
f
ind
(
i
)
−
(
,
(3)
where:
f
ind
(
i
)isthevalueofthefunctioncalculatedforthemodelforparam-
eterscorrespondingtotheindividual
i
exp
isanexperimentalvalue.
Theiterativesimulationofevolutionstepsisrelatedtotheselectionprocess.
IntheappliedAG,ingeneratedrandomlypopulationwechooserandomlytwo
individualsandtheonewithbetterfitnessgoestothenewpopulation.Thiskind
ofselectioniscalledthetournamentselection.
Thenewpopulationinthenextstepismutated.Themutationisanoperation
where a single bit of the individual from population is changed (or not) in a
randommanner.Intheconsideredcaseindividualsarebinaryencoded.Atypical
mutationoperationispresentedinFig8.
i
and
Fig. 8.
Exampleofbinarymutationofanindividual
Themutationprocessonnewpopulationindividualsisappliedwithacertain
probability,setup bythe user,thusnotallindividualsaremutated. Thenext
stepintheevolutionloopiscalledcrossover.Thisoperationinvolvesexchanging
Search WWH ::
Custom Search