20.500.12556/ReVIS-11461 Načrtovanje nevronske mreže za napovedovaje kvalitete tlačnega litja. Designing a neural network to predict die casting quality. Diplomska naloga raziskuje področje umetne inteligence, s poudarkom na uporabi nevronskih mrež za napovedovanje kakovosti pri visokotlačnem litju. Uvodnemu delu, ki podaja kontekstualni okvir, sledi teoretični del, ki obravnava ključne vidike umetne inteligence, vključno z zgodovinskim razvojem umetne inteligence in opisom strojnega učenja, ki vključuje nadzorovano, nenadzorovano in spodbujevalno učenje. Prav tako so v tem delu opisane nevronske mreže, njihova sestava, arhitekturne značilnosti ter lastnosti, skupaj s procesi modeliranja, določitvijo topologije in metrikami za ocenjevanje učinkovitosti regresijskih modelov. Del je zaključen s pregledom programskega jezika Python in njegovih knjižnic za strojno učenje, ki podpirajo analizo v empiričnem delu naloge. Empirični del naloge vključuje razvoj pilotnega modela napovedovanja kakovosti visokotlačnega litja s pomočjo orodja Orange ter implementacijo nevronske mreže v programskem jeziku Python. S pomočjo teh pristopov naloga prikazuje možno pot za napovedovanje kakovosti izdelkov pri visokotlačnem litju. The diploma thesis explores the field of artificial intelligence, with a focus on the use of neural networks for predicting quality in high-pressure die casting. The introductory section, which provides the contextual framework, is followed by a theoretical part that addresses key aspects of artificial intelligence, including the historical development of AI and a description of machine learning, encompassing supervised, unsupervised, and reinforcement learning. This section also describes neural networks, their structure, architectural features, and properties, along with modeling processes, topology determination, and metrics for evaluating the performance of regression models. The theoretical part concludes with an overview of the Python programming language and its machine learning libraries that support the analysis in the empirical part of the thesis. The empirical part of the thesis includes the development of a pilot model for predicting the quality of high-pressure die casting using the Orange tool and the implementation of a neural network in Python. Through these approaches, the thesis demonstrates a potential pathway for predicting product quality in high-pressure die casting. umetna inteligenca strojno učenje nevronske mreže Python visokotlačno litje. artificial intelligence machine learning neural networks Python high-pressure die casting. true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2025-02-23 16:52:24 2025-02-23 16:52:24 2025-02-25 13:03:11 0000-00-00 00:00:00 2025 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 13604 227277827 DIP_Ljimani_Dejhan_2025.pdf DIP_Ljimani_Dejhan_2025.pdf 1 B9CC7AB084C26A016A9354FC62D9C3E5 5c04a8c3c51ce8153bbb558ad73cb312d08adbb1228aa1ce2f15341ed48aa595 2d404567-f1fe-11ef-9f1e-001a4af901a5 https://revis.openscience.si/Dokument.php?lang=slv&id=13168 Univerza v Novem mestu - Fakulteta za ekonomijo in informatiko 0 0 0