»Diplomsko delo obravnava korekcijo izvorno-ciljne matrike v prometnih modelih z namenom izboljšanja ujemanja med modeliranimi in izmerjenimi prometnimi obremenitvami. V teoretičnem delu so predstavljene osnove štiristopenjskega prometnega modela, pomen izvorno-ciljne matrike in metode za njeno korekcijo. V praktičnem delu sta bili v programskem okolju PTV Visum uporabljeni metoda najmanjših kvadratov in metoda TFlowFuzzy. Za analizo občutljivosti je bila začetna matrika spremenjena za –5 %, +5 %, –10 %, +10 %, nato pa je bila za vsak scenarij izvedena korekcija. Uspešnost metod je bila ovrednotena s statističnim kazalnikom GEH in determinacijskim koeficientom R². Rezultati kažejo, da korekcija izvorno-ciljne matrike izboljša ujemanje modeliranih prometnih tokov s podatki prometnih štetij. Metoda najmanjših kvadratov je v obravnavanih scenarijih dosegla nekoliko boljše ujemanje s prometnimi štetji, medtem ko je analiza sprememb posameznih izvorno-ciljnih parov pri izbranem scenariju pokazala manjše spremembe pri metodi TFlowFuzzy.«
“The thesis deals with the correction of the source-target matrix in traffic models in order to improve the match between modelled and measured traffic loads. The theoretical part presents the basics of the four-stage traffic model, the meaning of the source-target matrix and the methods for its correction. In the practical part, the least squares method and the TFlowFuzzy method were used in the PTV Visum software environment. For the sensitivity analysis, the initial matrix was changed by −5%, +5%, −10% and +10%, and then a correction was performed for each scenario. The performance of the methods was evaluated using the GEH statistical indicator and the R² coefficient of determination. The results show that the correction of the source-target matrix improves the match of the modelled traffic flows with the traffic count data. The least squares method achieved a slightly better match with the traffic counts in the scenarios considered, while the analysis of changes in individual source-destination pairs in the selected scenario showed smaller changes in the TFlowFuzzy method.”
