20.500.12556/RUL-188197 Prilagajanje železniškega voznega reda zamudam magistrsko delo Railway timetable rescheduling due to delays Železniški promet je dovzeten za nepredvidene motnje in zamude, ki zahtevajo hitre operativne prilagoditve voznega reda. V magistrskem delu obravnavamo problem avtonomnega prilagajanja železniškega voznega reda z namenom minimizacije sekundarnih zamud v realnem času. Razvit je matematični model mešanega celoštevilskega linearnega programiranja, ki temelji na grafu alternativ ter dosledno upošteva varnostne in infrastrukturne omejitve. Zaradi računske zahtevnosti problema je implementirana metoda časovne dekompozicije z drsečim horizontom, ki omogoča hiter odziv sistema. Jedro rešitve predstavlja večkriterijska optimizacija, ki združuje leksikografsko metodo z iskanjem Paretove fronte. Z upoštevanjem realnih podatkov o številu potnikov sistem aktivno išče kompromise med robustnostjo prilagoditve voznega reda in zamudami potnikov. Predlagani model je implementiran v programskem jeziku Python z uporabo reševalnika SCIP ter ovrednoten na realnih primerih Slovenskih železnic za enotirno progo Ljubljana - Kamnik Graben v letu 2025. Rezultati stresnih testov in primerjava s kapacitetno analizo po metodi UIC 406 kažejo, da razviti sistem učinkovito razrešuje konflikte, omejuje širjenje zamud ter dispečerjem ponuja prilagodljive strategije za optimizacijo železniškega prometa. Railway traffic is susceptible to unforeseen disruptions and delays that require rapid operational timetable adjustments. In this master's thesis, we address the problem of autonomous railway timetable rescheduling with the aim of minimizing secondary delays in real time. A mixed-integer linear programming model has been developed, based on an alternative graph and consistently taking into account safety and infrastructure constraints. Due to the computational complexity of the problem, a rolling horizon time decomposition method has been implemented, enabling a rapid system response. The core of the solution is a multi-objective optimization approach that combines the lexicographic method with Pareto frontier search. By incorporating real passenger count data, the system actively seeks trade-offs between the robustness of timetable rescheduling and passenger delays. The proposed model has been implemented in the Python programming language using the SCIP solver and evaluated on real-world examples from Slovenian Railways for the single-track Ljubljana -- Kamnik Graben railway line in 2025. The results of stress tests and a comparison with the UIC 406 capacity analysis method show that the developed system effectively resolves conflicts, limits the propagation of delays, and provides dispatchers with flexible strategies for railway traffic optimization. železniški promet vozni red zamude optimizacija MCLP drseči horizont večkriterijska optimizacija SCIP railway traffic timetable delays optimization MILP rolling horizon multi-objective optimization SCIP true false false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2026-09-19 08:15:43 2026-09-19 08:15:50 2026-09-25 10:32:05 0000-00-00 00:00:00 2026 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 519.8 164615 291467779 24159.pdf 24159.pdf 1 DDE35BF0E8302747AABB61931A1BC185 fd7a364c74d7381331f253254e16c3a2bc72bc070e37c0e36e9833549241d7ff 81f20246-b3f1-11f1-8bc5-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=248226 Fakulteta za matematiko in fiziko 0 0 0