Fault diagnosis for energy conversion systems
No Thumbnail Available
Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Université Badji Mokhtar Annaba
Abstract
This thesis delves into the complex dynamics of energy conversion systems, with afocused exploration of induction motors due to their crucial role in various industrial applications. Central to this study is the enhancement of fault diagnosis techniques aimed at improving the reliability and efficiency of these motors. Employing advanced signal processing and artificial intelligence, the research innovatively applies Continuous Wavelet Transform to analyse vibration signals for precise identification of fault characteristics in time-frequency representations. This method highlights subtle anomalies often missed by traditional diagnostic approaches. Additionally, the thesis utilises Convolutional Neural Networks, specifically the efficient SqueezeNet architecture, to classify these faults based on features extracted from scalograms. This integration of sophisticated signal processing with cutting-edge machine learning technologies provides robust solutions for the real-time detection and classification of induction motor faults, thereby enhancing system reliability and operational efficiency. By advancing diagnostic capabilities, this research contributes significantly to the sustainability and efficiency of energy conversion systems, positioning these advanced diagnostic technologies at the forefront of industrial applications and maintenance strategies. The findings promise to redefine maintenance protocols and improve the longevity and performance of energy systems, ensuring they meet modern demands for energy efficiency and reliability.
Description
Keywords
energy Conversion Systems; induction Motors; fault diagnosis; continuous wavelet transform; convolutional neural networks