20.500.12556/DKUM-99153 Uvajanje Azure Ai asistenta za napovedovanje okvar strojev Implementation of an Azure Ai assistant for predictive maintenance Nenačrtovane zaustavitve CNC-strojev povzročajo izgubo časa, višje stroške vzdrževanja in zmanjšano učinkovitost proizvodnje. V magistrskem delu je bil razvit Azure AI-asistent za podporo napovedovanju okvar. Rešitev vključuje model strojnega učenja v Azure Machine Learning, Azure Function, OpenAPI povezavo in uporabo pristopa RAG za tehnično dokumentacijo. Model je dosegel skupno točnost 98,66 %, vendar je zaznal približno 66,4 % dejanskih okvar in spregledal 114 okvar. Sistem je zato obravnavan kot demonstracija prenosljive arhitekture, ne kot neposredno potrjena rešitev za realni CNC-stroj. Unplanned CNC machine downtime results in lost time, higher maintenance costs, and reduced production efficiency. This master's thesis involved the development of an Azure AI assistant designed to support failure prediction. The solution incorporates a machine learning model within Azure Machine Learning, an Azure Function, an OpenAPI connection, and a Retrieval-Augmented Generation (RAG) approach applied to technical documentation. Although the model achieved an overall accuracy of 98.66%, it detected approximately 66.4% of actual failures and missed 114 failure instances. Consequently, the system is regarded as a demonstration of a transferable architecture rather than a fully validated solution for a real-world CNC machine. umetna inteligenca Microsoft Azure AI asistent CNC stroj artificial intelligence Microsoft Azure AI assistant CNC machine true true false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2026-08-04 16:23:59 2026-09-03 10:46:30 2026-09-04 03:08:40 0000-00-00 00:00:00 2026 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 MAG_Petek_Regorsek_Aljaz_2026.pdf MAG_Petek_Regorsek_Aljaz_2026.pdf 1 57C28B5F5EE291C7EA2D315292707DC8 69dcb36af9ddb5dbedc1d887ce73be336d57b6c96a14ee115dcf9f46eac35e78 82f56750-a6b0-11f1-aef5-00155d0001be https://dk.um.si/Dokument.php?lang=slv&id=211275 Fakulteta za strojništvo 0 0 0