IMPROVING SYNCHRONOUS MOTOR MODELLING
WITH ARTIFICIAL INTELLIGENCE
Petar Čisar1, 2
1University of Criminal Investigation and Police Studies Received: 22nd May 2024. ABSTRACT Synchronous motors are essential in various industrial and commercial applications because of their efficiency and
constant speed operation. Accurate modelling of these motors is crucial for optimizing performance, control, and maintenance. Traditional modelling
methods, such as the d-q reference frame method, often fall short in terms of complexity and accuracy, especially under dynamic conditions. This study
aims to enhance synchronous motor modelling using machine learning algorithms, specifically focussing on predicting the excitation current, a critical
parameter for motor performance. KEY WORDS CLASSIFICATION,
Sanja Maravić Čisar3
and
Attila Pásztor2
Belgrade, Serbia
2John Von Neumann University, GAMF Faculty of Engineering and Computer Science
Kecskemét, Hungary
3Subotica Tech-College of Applied Sciences
Subotica, Serbia
INDECS 22(3), 329-340, 2024
DOI 10.7906/indecs.22.3.8
Full text available in
pdf format.
Accepted: 16th June 2024.
Regular article
In this research, a dataset comprising synchronous motor operational parameters was analysed using various machine learning techniques. The primary
methods evaluated include regression and M5 algorithms. The evaluation criteria were the time required to build and test the models and the accuracy of
their predictions. Our findings indicate that both the regression and M5 algorithms significantly outperform traditional methods, providing more precise
and efficient models for synchronous motor behaviour under diverse operating conditions.
synchronous motors, parameters, machine learning, prediction, excitation current
ACM: I.2.6, I.5.1
JEL: C45, C63
PACS: 07.05.Mh