Treść książki
Przejdź do opcji czytnikaPrzejdź do nawigacjiPrzejdź do informacjiPrzejdź do stopki
JerzyTchórzewski,TomaszKania
tochoosetherightmethodofdataanalysis,which,asnotedin[27],largely
determinesthequalityoftheobtainedresearchresults.Thedataasthe
resultsofempiricalactivityoccurringinrealityweretakenfromthewebsite
ofPolskieSieciElektroenergetyczneS.A.(PSES.A.,PSE)[22,32],includ-
ing,interalia:
−dateandtimeofmeasurement,
−domesticelectricitydemand[MW],
−nationalbalanceofelectricityexchange(ee)inparallelinterconnec-
tion,
−nationalbalanceofnon-paralleleeexchange[MWh]andothervalues.
Similarintermsofdomain,researchondataanalysiswaspreviously
carriedoutattheWrocławUniversityofTechnologyinthefieldofassess-
mentofvariousoperatingstatesofpowerminingnetworksintermsofelec-
tricityquality,theresultsofwhichwerepublishedin[5].There,hierarchical
aswellasnon-hierarchicalmethodswereusedforclusteranalysis.
Intheexperimentconsideredinthisstudy,theappropriateclusteranal-
ysiswasprecededbyapreliminarydataanalysisinordertoidentifyand
evaluatevariousoperatingstatesofthetestedPPSsystem[8,12-13,27]
withtheuseofMATLABandSimulink[30].Suchanapproachisdictated
bytherightobservationofR.Tadeusiewiczthatthemethodsofcluster
analysisare,inasense,anoppositeissueinrelationtothemethodsof
imagerecognition[18].
Thisisduetothefactthatwhenrecognizingimages,thereareprede-
terminedcategories,andacertainalgorithm(e.g.patternrecognition)isto
determineforanewobjecttowhichcategoryitcanbeclassified,andin
theclusteranalysistherearedataofmanyobjects,andacertainalgorithm
(e.g.clusteranalysis)istodeterminetohowmanyclusters(category
equivalents)theseobjectscanbeassignedanddividedamongthede-
tectedclusters.
Therearemanydefinitionsoftheconceptofclusteranalysis,oneof
themisthedefinitiongivenin[30]:"Clusteranalysisconsistsindistinguish-
inggroupscalledclustersinthesetofanalyzeddata".Itwasassumed
that:"clustersaredisjointsubsetsofadatasetwithsuchapropertythat
databelongingtodifferentclustersdiffersmuchmorefromeachotherthan
databelongingtoacommoncluster".Thus,thetaskofclusteranalysisis
todiscovertheexistingregularitiesinthedataset,andthemeansofde-
tectingthemissomemeasureofsimilarityordissimilarity.Contemporary
methodsofdataanalysisarebasedmainlyontheuseofcomputersto
12
