Abstract:

Fault diagnosis in electrical machines is a cornerstone of operational reliability and cost-effective maintenance strategies. This review provides a comprehensive exploration of the integration of higher-order spectral analysis (HOSA) techniques—such as a bispectrum, spectral kurtosis, and multifractal wavelet analysis—with advanced artificial intelligence (AI) methodologies, including deep learning, clustering algorithms, Transformer models, and transfer learning. The synergy between HOSA’s robustness in noisy and transient environments and AI’s automation of complex classifications has significantly advanced fault diagnosis in synchronous and DC motors. The novelty of this work lies in its detailed examination of the latest AI advancements, and the hybrid framework combining HOSA-derived features with AI techniques. The proposed approaches address challenges such as computational efficiency and scalability for industrial-scale applications, while offering innovative solutions for predictive maintenance. By leveraging these hybrid methodologies, the work charts a transformative path for improving the reliability and adaptability of industrial-grade electrical machine systems.

Aplicación:

Esta investigación resulta especialmente útil para la huerta valenciana, donde el uso de bombas, motores y sistemas eléctricos es esencial para el riego y otros procesos agrícolas. La aplicación de estas tecnologías permite detectar de forma temprana fallos como cortocircuitos, desgaste de rodamientos o desequilibrios en los motores, lo que reduce paradas imprevistas y costes de mantenimiento. Así, se mejora la fiabilidad y eficiencia de los sistemas eléctricos agrícolas, favoreciendo una agricultura más sostenible y tecnológicamente avanzada.Â

We would like to acknowledge funding from the Generalitat Valenciana (Spain) through the PROMETEO 2024 CIPROM/2023/32 grant.