Estimating Electric Motor Temperatures With Machine Learning Models

Abstract

The master's thesis was completed by Pastukh Yaroslav Tarasovych, a student of the KNSS-22 group. Master’s degree work of the student of the group CSAI-12. The topic is "Estimating Electric Motor Temperatures With Machine Learning Models". The work is aimed at obtaining a master's degree in 122 "Computer Science". The object of research is the program for predicting temperature of the electric motor. The subject of research is methods and algorithms for building hybrid systems of computational intelligence. The goal of the research is achieved by training a regression model on the input data. Further processing of the dataset is performed using classical methods of data preprocessing and data scaling. The modeling of the electric motor temperature is implemented using various machine learning and deep learning algorithms based on real dataset collected by manufacturers of the selected type of electric motor to solve the task of real-time temperature prediction. As a result of the master's qualification work, a program for predicting the temperature of several components of the electric motor in real time based on input data has been developed, which allows to investigate the performance of various machine learning methods, select hyperparameters for optimal system operation, and compare training results on the test sample. The total volume of work: 53 pages, 28 figures, 21 references.