Locally optimized chirplet spectrogram for condition monitoring of induction machines in transient regime

Martínez-Román, Javier; Puche-Panadero, Rubén; Sapena-Bañó, Ángel; Burriel-Valencia, Jordi; Riera-Guasp, M.; Pineda-Sánchez, Manuel

doi:10.1016/j.measurement.2021.110690

Cited by 4 publications

(1 citation statement)

References 91 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, an optimized stationary wavelet‐packet transform‐based feature is extracted from the motor current signature under various bearing failure conditions. Martinez‐Roman et al (2022) developed a locally optimized chirplet spectrogram that could provide an affordable, high‐resolution spectrogram signature to help identify the fault harmonics. Atta et al (2021) discussed a hybrid machine‐learning approach for detecting induction‐motor faults that use motor parameters like stator currents and vibration signals to reveal significant information about the state of the motor.…”

Section: Impact and Related Workmentioning

confidence: 99%

Nature‐inspired artificial bee colony‐based hyperparameter optimization of CNN for anomaly detection in induction motor

et al. 2023

View full text Add to dashboard Cite

The Induction Motor (IM) is one of the most frequently used prime movers in most industrial and transportation systems. The motor's stable and safe operation directly influences the secure and reliable operation of such prime movers. Developing an intelligent fault diagnosis system for such motors is very significant. This paper presents an intelligent fault diagnosis method based on the improved functionality of a Convolutional Neural Network (CNN) through its hyperparameter optimization using a nature‐inspired Artificial Bee Colony Optimization (ABCO) algorithm. The proposed diagnostic method introduces and analyses various possible mechanical and electrical faults in the IM. The validation of the proposed method is presented with three different modalities, including vibration, acoustic, and infrared thermography, with their comparative performance analysis. Vibration and acoustic‐based detection are done with time‐frequency scalograms using Constant Q Transform (CQT), which provides enhanced time resolution for lower and higher frequencies. The obtained result indicates that the infrared thermography‐based anomaly detection outperforms the vibration and acoustic‐based diagnosis with 100% classification accuracy. The results signify the potential to diagnose different mechanical and electrical faults in IM with substantial reliability and robustness.

show abstract

Section: Impact and Related Workmentioning

confidence: 99%