Zaprtozančno krmiljenje hitrosti vrtenja elektromotorja s spremenljivo obremenitvijo na osnovi vzpodbujevalnega učenja

Komar, Žiga (author), Kozjek, Dominik (mentor)

Diplomska naloga obravnava problem zaprtozančnega krmiljenja hitrosti vrtenja enosmernega elektromotorja s spremenljivo obremenitvijo z uporabo metode vzpodbujevalnega učenja. Klasične metode, kot je PID krmiljenje, pogosto ne zagotavljajo optimalnih rezultatov pri nelinearnih in dinamično spreminjajočih se sistemih, zato je bila v nalogi raziskana možnost uporabe umetne inteligence. Ključni del naloge je bilo ocenjevanje vpliva različnih kombinacij opazovanj na učinkovitost učenja in delovanja krmilnika.

• Language: Slovenian
• Keywords: elektromotor, krmiljenje, strojno učenje, vzpodbujevalno učenje
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FS - Faculty of Mechanical Engineering
• Year: 2025
• Number of pages: X, [1] f. pril. 30 f.
• PID: 20.500.12556/RUL-168247
• UDC: 681.516.3:621.313.2:004.85(043.2)
• Publication date in RUL: 05.04.2025

Secondary language

• Language: English
• Title: Closed-loop speed control of a variable load electric motor based on reinforcement learning

Abstract

The thesis addresses the problem of closed-loop speed control of a DC motor with variable load using the reinforcement learning method. Classical methods, such as PID control, often fail to provide optimal results in nonlinear and dynamically changing systems. Therefore, the thesis explores the potential application of artificial intelligence. A key part of the research focuses on evaluating the impact of different combinations of observations on the learning efficiency and performance of the controller.

• Keywords: electric motor, control, machine learning, reinforcement learning