Feature Extraction in a Diesel locomotive Starter Motor with Wavelet Transform method

Karimi, Amirhosein; Ghorbani Vaghei, Bahman; Radkani, Mahsa; Pasokhi, Husain; Masomi, Zahra

doi:10.22068/ijrare.333

Volume 10, Issue 2 (12-2023) IJRARE 2023, 10(2): 19-28 | Back to browse issues page

‎ 10.22068/ijrare.333

Mendeley

Zotero

RefWorks

Karimi A, Ghorbani Vaghei B, Radkani M, Pasokhi H, Masomi Z. Feature Extraction in a Diesel locomotive Starter Motor with Wavelet Transform method. IJRARE 2023; 10 (2) :19-28
URL: http://ijrare.iust.ac.ir/article-1-333-en.html

Feature Extraction in a Diesel locomotive Starter Motor with Wavelet Transform method

Amirhosein Karimi

, Bahman Ghorbani Vaghei

, Mahsa Radkani

, Husain Pasokhi

, Zahra Masomi

School of Railway Engineering, Iran University of Science and Technology

Abstract: (618 Views)

Nowadays, due to the increasing growth of the railway industry and the need to increase safety in trains, component failure must be predicted and identified before it occurs. The direct current motor is one of the most important pieces of equipment in today's industries, especially in the rail transportation industry, and is used in various parts of electric and diesel electric locomotives, such as traction motors, train starters, cooling systems, snow wipers, etc. This article aims to categorize healthy and faulty motors and provide a method to separate them. For this purpose, a number of locomotive starter motor starting current data points have been used. The collected data, half of which are healthy and half of which are defective, are decomposed into nine levels by discrete wavelet transform with the Debuchies4 function, and the detail and estimation coefficients of each data point are calculated by MATLAB software. Unique features of the data were identified to represent it and separate the healthy and defective classes. Before the wavelet transformation, skewness, kurtosis, and root mean square parameters were extracted from all the data. The features showed that the data were intertwined, making it difficult to separate them, and a more complex classifier was needed. However, after applying wavelet transformation, the data was separated at different levels and could be separated from each other with a simple linear classifier.

Keywords: Locomotive, DC Motor, Feature Extraction, Wavelet Transform, Fault Detection

Full-Text [PDF 564 kb] (191 Downloads)

Type of Study: Research | Subject: Electrical railway